JP5189906B2 - Method and apparatus for detecting abnormal accumulation of bacteria, and graphing method and apparatus for accumulating warning scores for detection of abnormal accumulation of bacteria - Google Patents

Description

  The present invention relates to a method and apparatus for detecting abnormal accumulation of bacteria, an anti-biogram classification method and apparatus, a method and apparatus for creating a two-dimensional carrier map, a method and apparatus for evaluating infection control indicators, and a graph of cumulative warning scores for detection of abnormal accumulation of bacteria. It relates to a method and a device.

  With the advancement of medical care and the aging of the population, the proportion of vulnerable patients with weak immunity in hospitals is increasing. Infectious patients also develop infectious diseases due to weakly pathogenic bacteria (so-called attenuated bacteria) that healthy individuals do not develop infectious diseases. Such an infection is called an opportunistic (hiyomi) infection, and an attenuated bacterium causing an opportunistic infection is called an opportunistic infection. Most of opportunistic bacteria are resident bacteria and environmental bacteria, exist in the resident bacterial flora such as the intestinal tract of patients and healthcare workers, and in the hospital environment, and stay in the hospital for a long time. In hospitals, antibacterial drugs (generic name for antibiotics made by living organisms and synthesized synthetic antibacterial drugs) are frequently used. Sensitive bacteria that work with antibacterial drugs are selected, and resistant bacteria that do not work with antibacterial drugs are selected. For this reason, many opportunistic infectious bacteria have become multidrug resistant and resist treatment with antibacterial agents, making treatment difficult when nosocomial infections due to these bacteria occur.

[Background on the necessity for detecting abnormal accumulation of bacteria]
Many hospital-acquired infections are intractable infections caused by such multidrug-resistant opportunistic bacteria. Today, methicillin-resistant staphylococci (MRSA), multi-drug resistant Pseudomonas aeruginosa (MDRP), etc. are in trouble, but since various resident bacteria and environmental bacteria cause nosocomial infections, those few Focusing on specific bacteria will overlook nosocomial infections and signs of other bacteria. In order to monitor the isolation status of all bacteria and grasp the signs of nosocomial infections, it is necessary to evaluate the isolation by collecting the isolation status of all bacterial species by ward or ward every day or at least every few days. There is. This requires enormous work and is difficult to do manually. Therefore, realization of a technique capable of automatically detecting abnormal accumulation of bacteria is strongly desired.

  In other words, nosocomial infections occur when the above-mentioned various bacteria are artificially spread in the hospital and the diffused bacteria cause infection, but when the bacteria are artificially spread in the hospital, they are spread. If the increase in the number of isolated bacteria, that is, abnormal accumulation of bacteria can be detected, the spread of artificial bacteria, the first stage of nosocomial infection, can be detected. It becomes possible to prevent nosocomial infections. In this regard, if real-time surveillance using a computer can be performed, it is possible to calculate the number of separations and separation rates of specific bacteria in real time, but it is possible to automate while maintaining the sensitivity and specificity of detection. There was no statistical method in the past.

  Therefore, the inventors of the present application have already invented detection of abnormal accumulation of bacteria using binomial distribution (see Non-Patent Documents 1 and 2) and solved this problem, but the probability of the null hypothesis. There are still points to be resolved such as what to use as the baseline rate (for example, how to determine the aggregation period and deduplication method), what period width and how often it is appropriate to aggregate. .

[Background on the need for antibiogram classification]
Nosocomial infections caused by bacteria are mainly caused by the causative bacteria spreading in the hospital. On the other hand, if the infection does not occur even if the bacteria spread, it will not become a nosocomial infection. If it is possible to detect and confirm the spread of bacteria at an early stage, hospital infections can be prevented. Even if abnormal bacterial accumulation can be detected by detecting the abnormal accumulation of bacteria, whether the accumulation is based on the spread of bacteria in the hospital, or a person who has the bacteria happens to be in the ward It is necessary to judge whether it gathered in. Bacteria are identified as the same species (for example, Pseudomonas aeruginosa) (ie, the species is determined as a result of the test), but each has a slightly different genetic background. Even if the human being is the same human being, it is the same as each person's gene being slightly different. The difference between humans and bacteria is that when humans inherit half of the genes from their father and mother when they proliferate, bacteria multiply by asexual division, so the original and proliferated bacteria However, it has the exact same gene except for the change caused by the rare mutation. Therefore, in the case of bacteria, it is possible to accurately predict whether the bacteria isolated at different locations and at different times are the same bacteria. The most accurate method for predicting whether bacteria isolated at different locations and times are identical is to determine the entire gene sequence of each bacterium. However, in reality, even if the entire gene sequence is not determined, the outline of the gene sequence is examined, or the characteristics of the bacteria (that is, traits such as skin color and hair color in humans) It can be predicted with considerable accuracy whether they are the same or different. This is due to the fact that bacteria generally rarely produce significant mutations in the short term (eg, in the order of tens of days), but in the long term they produce various mutations, including large-scale gene recombination.

  The most commonly used feature of such bacteria in clinical settings is the so-called antibiogram, which is a pattern of resistance to different types of antibacterial drugs (ie, the ineffectiveness of the drug). If one bacterium is resistant to the antibacterial drug ABC and sensitive to DEF (ie, the drug works), and another bacterium belonging to the same species is resistant to the antibacterial drug ABCD and sensitive to EF, It can be predicted that it has a different origin. If you select an appropriate combination of drugs, all combinations (ie patterns) such as resistance to ACE and sensitivity to BDF may appear, and if the antibiogram is the same, the origin of those bacteria may be the same It can be concluded that at least cannot be denied. This method is based on such epidemiology because the susceptibility testing of bacteria to antibacterial drugs (that is, whether they are resistant or sensitive) is within the scope of insurance practice and is widely conducted as a daily clinical test. It has the advantage of not incurring new costs for research (ie, research methods to clarify causal relationships between diseases that occur in the population, such as infectious diseases and diseases caused by environmental factors). It is known that the antibiogram prediction of bacterial identity is in good agreement with the results of classification based on the outline of gene sequences and detailed gene analysis (these analyzes are called molecular epidemiological methods). Some of the antibiograms predicted to be identifiable by the antibiogram include those that are not the same, but the bacteria that were determined to be different by the antibiogram were identified as different by other methods. This is the first survey to be conducted, judging whether detailed molecular epidemiological methods should be performed. In general hospitals, analysis by molecular epidemiology is almost impossible in terms of both equipment and cost. Only in cases where many health hazards have occurred, universities, prefectural health laboratories, national infectious disease laboratories, etc. conduct molecular epidemiological studies.

  Currently, antibiograms are manually organized based on the susceptibility of bacteria to multiple antimicrobial agents obtained as a result of clinical tests. At this time, (1) the susceptibility of the bacteria is so-called SIR determination as S (sensitivity), I (intermediate), and R (resistance). The difference between I and S and I and R may be an error range of inspection. (2) When organizing the antibiograms of a large number of bacteria, not all the susceptibility of the same antibacterial drug is examined for all bacteria. (3) There is no single grouping. There are problems such as that it takes a lot of time to work, and the results are not confident.

  Therefore, many clinical technologists who perform bacteriological tests hope to realize technology that automates the organization of antibiograms.

  Antibiogram automatic organization is a classification of patterns, and because it was thought that it contained ambiguous elements for the above reasons, there was a mechanism to create an antibiogram using a computer, No program has been implemented for classification.

[Background on the need for 2D carrier maps]
It is also an important technique to color-code the antibacterial agent classification based on the antibiogram classification and create a so-called two-dimensional carrier map, and it is considered that the diffusion state of the bacteria can be easily grasped.

  Although the two-dimensional map of the isolation situation of a specific bacterium has been performed so far, there was no method of antibiogram classification as described above. It was not possible to automatically create a material for analysis in consideration.

  Therefore, it is also useful to realize a technique for automatically creating a two-dimensional carrier map in combination with such an antibiogram automatic classification.

[Background on the necessity for evaluation of PID infection index]
In hospital infection countermeasures, it is important to graph various indicators and examine countermeasures based on values and fluctuations. For example, if the number of isolated MRSA (methicillin-resistant Staphylococcus aureus) MRs (methicillin-resistant Staphylococcus aureus), which is often a problem as a hospital-acquired bacterium, is graphed, and they tend to increase, call attention to the person in charge. In order to eliminate the cause of diffusion in the inside, the contact preventive measures will be strengthened. In general, changes are clarified by graphing, but it is not always easy to accurately grasp the trend of many indicators by visual observation. (1) Slight deviation from the target value, (2) Speed of change (Slope of the graph) is not always accurately grasped.

  PID (proportional-integral-differential) control is a method that has been used for a long time for temperature and speed control, but it has become more practical with the development of computer technology, and is widely used for consumer and industrial control. ing.

  Therefore, according to the PID technology, it is possible to accurately evaluate the rate of change, the continuation of a slight deviation, and the deviation, and the amount of difference in the change rate and the duration of the minute deviation that is sometimes overlooked by visual inspection. Focusing on the fact that it can be converted to a graph, keeping various indicators in infection countermeasures normal is considered as one control, and it is considered useful to apply PID control to trend analysis of infection countermeasures.

[Background on Necessity for Graphing Accumulation of Detection Warning Score of Abnormal Accumulation of Fungi]
Using the above-mentioned detection technique for abnormal accumulation of bacteria using the binomial distribution and accumulating the warning score obtained as an output may be useful for measures against hospital infections. It has not been.
Shuhei Fujimoto, “Small and Medium-sized Hospital Infectious Disease Monitoring System”, Health and Labor Sciences Research Grants Medical Safety Comprehensive Research Project, Medical Safety Comprehensive Research Project Establishing a monitoring system to prevent nosocomial infections and improving the function of the bacterial laboratory Research 2005 Report, 13-102. (2006) Shuhei Fujimoto, “Study on Efficient In-Hospital Infection Control Surveillance”, Grants-in-Aid for Scientific Research on Welfare Science, Research on Emerging / Re-emerging Infectious Diseases Research Network on Trends in Development of Drug-resistant Bacteria 2003 Summary Research Report, 87-94. (2004)

  In view of the circumstances as described above, the present invention is capable of automatically detecting an abnormal accumulation of statistically significant bacteria from the results of a bacterial test using a computer system. It is an object of the present invention to provide a method and an apparatus for detecting abnormal accumulation of bacteria.

  Further, the present invention provides an anti-biogram classification method and an anti-biogram classification apparatus that can save labor and prevent errors and oversights by automating the organization of the antibiograms, that is, the classification of patterns, to reduce the epidemiological survey of nosocomial infections. It is an issue to provide.

  In addition, the present invention is color-coded antibacterial classification based on antibiogram classification, plotting bacteria separation on the time axis and space axis, connecting the same patient with a straight line, and bacteria with the same resistance pattern diffuse in the ward etc. It is an object of the present invention to provide a two-dimensional carrier map creation method and a two-dimensional carrier map creation device that can automatically create a two-dimensional carrier map that can easily grasp the current state.

  In addition, the present invention applies PID control to trend analysis of infection countermeasures, can keep various indicators in infection countermeasures normal, and it is inappropriate to use real numbers as indicators, or when the integration period is long Eliminates the failure of an improperly long period of influence in the past, and makes it possible to accurately perform various graph evaluations, that is, a graph evaluation technique based on the PID method, that is, an evaluation method and apparatus for an infection control index based on the PID method It is an issue to provide.

  Further, the present invention provides a graphing method and apparatus for accumulating warning scores for automatic accumulation detection of bacteria, which is capable of accumulating and graphing warning scores obtained as outputs using the abnormal accumulation detection of the bacteria. The issue is to provide.

The present invention is intended to solve the above-described problems.
For each place to be detected for abnormal accumulation of bacteria, the bacteria were positive for each bacteria for the specimen submitting patient and all the bacteria to be detected during the predetermined observation period at the specified frequency. Count the number of patients,
For all bacteria, using a binomial distribution based on the total number of positive patients for each bacterium, the number of patients submitting specimens, and the baseline rate, which is the positive rate for each bacterium determined in advance, The probability of Bernoulli trials to be established is calculated according to the null hypothesis that there is no bias in bacterial isolation and that bacterial isolation is controlled only by chance,
When the calculated probability is smaller than the default value, the probability of being controlled only by chance is small, and it is determined that there is an abnormal accumulation due to bias in the separation of bacteria,
A method for detecting abnormal accumulation of bacteria,
When calculating the baseline rate, by deduplicating the same patient so as to be the total number of months and people, the baseline rate is determined as the probability of a single independent trial of the binomial distribution. To be,
A method for detecting abnormal accumulation of bacteria,
Also provide
For each place to be detected for abnormal accumulation of bacteria, the bacteria were positive for each bacteria for the specimen submitting patient and all the bacteria to be detected during the predetermined observation period at the specified frequency. Total data calculated in advance by counting the number of patients,
For all bacteria, using a binomial distribution based on the total number of positive patients for each bacterium, the number of patients submitting specimens, and the baseline rate , which is the positive rate for each bacterium determined in advance, The probability of Bernoulli trials to be established is calculated according to the null hypothesis that there is no bias in bacterial isolation and that bacterial isolation is controlled only by chance,
When the calculated probability is smaller than the default value, the probability of being controlled only by chance is small, and it is determined that there is an abnormal accumulation due to bias in the separation of bacteria,
A device for detecting abnormal accumulation of bacteria,
When calculating the baseline rate, by deduplicating the same patient so as to be the total number of months and people, the baseline rate is determined as the probability of a single independent trial of the binomial distribution. To be,
An apparatus for detecting abnormal accumulation of bacteria is provided.

In addition, the invention of the present application is to solve the above problems.
A method for classifying an anti-biogram, which is a drug resistance pattern possessed by bacteria, using SIR determination,
Among the combinations of two bacteria, a combination having at least one difference between S and R for all antibacterial drugs is determined as a mismatched combination, and the number of drugs that match between the two bacteria is defined as 0 points. , Determining the combination that is S or R to each other as a matching combination, and determining the number N of drugs that match between the two bacteria,
This match / mismatch is determined for all combinations of bacteria,
Create a group of bacteria whose number N of matching drugs is greater than or equal to a predetermined threshold,
Anti-biogram classification method, characterized by
Also provide
A device for classifying an anti-biogram, which is a drug resistance pattern of a bacterium, using SIR determination,
Among the combinations of two bacteria, a combination having at least one difference between S and R for all antibacterial drugs is determined as a mismatched combination, and the number of drugs that match between the two bacteria is defined as 0 points. , A means for determining a combination that is S or R to each other as a matching combination, and determining the number N of drugs that match between the two bacteria,
Means for determining this match / mismatch for all combinations of bacteria,
Means for creating a group of bacteria in which the number N of matching drugs exceeds a predetermined threshold;
An anti-biogram classification device, comprising:
I will provide a.

In addition, the invention of the present application is to solve the above problems.
A method of creating a two-dimensional carrier map based on a group of bacteria classified by the anti-biogram classification method,
Color-code the group of bacteria,
Based on the three parameters of separated patient, location, and time, plot the separation data of each bacterium along the time axis and space axis on the two-dimensional carrier map,
Connecting the same patient with a line,
A two-dimensional carrier map creation method characterized by
Also provide
A device for creating a two-dimensional carrier map based on a group of bacteria classified by the anti-biogram classification device,
Color code means for the group of bacteria;
Means for plotting the separation data of each bacterium along the time axis and the space axis on the two-dimensional carrier map based on the three parameters of the separated patient, location and time;
Means for connecting the same patient with a line;
A two-dimensional carrier map creation device, comprising:
I will provide a.

In addition, the invention of the present application is to solve the above problems.
A method for evaluating fluctuations and biases in numerical values that are indicators of infection control,
For each predetermined observation period unit, P factor indicating the difference from the target value, D factor indicating the difference from the previous observation unit on the time axis, and I factor indicating the sum of the deviations of the past predetermined period are calculated. And
The sum of these P factor, D factor, and I factor is used as a PID value, and the PID value or a positive PID value that is the absolute value of the PID value is used as an index of the degree of effort to reach the target value of the infection control result. ,
A method for evaluating an infection control index characterized by
Also provide
A device that evaluates fluctuations and biases in numerical values that are indicators of infection control,
For each predetermined observation period unit, P factor indicating the difference from the target value, D factor indicating the difference from the previous observation unit on the time axis, and I factor indicating the sum of the deviations of the past predetermined period are calculated. Means to
The sum of these P factor, D factor, and I factor is used as a PID value, and the PID value or a positive PID value that is the absolute value of the PID value is used as an index of the degree of effort to reach the target value of the infection control result. Means,
An infection countermeasure index evaluation apparatus characterized by comprising:
I will provide a.

In addition, the invention of the present application is to solve the above problems.
It is a method of accumulating the warning score obtained as an output of the abnormal accumulation detection method of the above-mentioned bacteria and graphing it,
The probability of Bernoulli trials representing the calculated separation of bacteria without bias is classified by a threshold, and the warning score obtained by giving a high score to those with low probability is tabulated for each bacterium for a certain period, and the counting result Graph with the horizontal axis as the time axis and the vertical axis as the warning score.
A graphing method for accumulating warning scores for detection of abnormal bacterial accumulation,
Also provide
An apparatus for accumulating and graphing a warning score obtained as an output of the above abnormal accumulation detection apparatus for bacteria,
Classifying the calculated probability of Bernoulli trials representing a non-biased separation of the bacteria by a threshold, and a warning score obtained by giving a high score to a low probability, a means for aggregating the bacteria every fixed period, Means for graphing the aggregated results with the horizontal axis as a time axis and the vertical axis as a warning score;
A graphing apparatus for accumulating warning scores for detecting abnormal accumulation of bacteria characterized by comprising:
I will provide a.

  Furthermore, the present invention also provides a program for causing a computer to execute the functions of the various methods or apparatuses, and a computer-readable recording medium on which the program is recorded.

[Detection of abnormal accumulation of bacteria]
The principle of detecting abnormal accumulation of bacteria in the present invention is as follows.

  Bacterial separation takes only two values, culture positive and negative, and when it is separated evenly, so-called sporadic separation, it happens by chance according to the baseline rate (baseline rate) independently each time. This is a so-called Bernoulli trial that occurs under the rule of the country. Pay attention to the Bernoulli trial without bias, the probability can be calculated by binomial distribution (for example, cumulative binomial distribution expressed by the following formula 1), there is no bias in the separation of bacteria, and the separation of bacteria is controlled only by chance If the calculated probability is sufficiently small and the calculated probability is sufficiently small, the probability of being controlled only by chance is small and there is some bias, that is, there is abnormal accumulation in the separation of bacteria I conclude that there was. As a result, if it is possible to evaluate whether or not there is a bias in the separation of bacteria for the entire hospital, ward, and ward for all bacteria, it is possible to automatically and reliably capture abnormal accumulation of bacteria for all species. become.

Here, p = baseline rate, n = number of specimen-submitted patients, and k = number of positive patients.
If a species is isolated from more patients than by chance at a particular location (such as a ward), it is suspected that spread (infection or colonization) has occurred in that location, and the actual spread (infection or infection) When (carrying) has occurred, (1) it is suggested that there may be a problem with the infection countermeasure (hygienic technique) at that location. (2) If spread continues, it is possible that it will lead to the occurrence (outbreak) of nosocomial infections with a certain probability. Therefore, if it is possible to automatically and reliably capture abnormal accumulation of bacteria for all bacterial species, outbreaks based on insufficient infection countermeasures can be prevented.

The detection of abnormal accumulation of bacteria using this binomial distribution is more specifically executed under the following setting conditions.
(1) When calculating the baseline rate, by deduplicating the same patient so that the total number of months and people is the same, the baseline rate can be calculated for a single independent trial of the binomial distribution. Try to be a probability. At this time, deduplication of the same patient may be performed every predetermined period, for example, every year. Further alternatively, the positive rate of the bacteria at the place or the national average positive rate is taken as the baseline rate, thereby realizing an objective evaluation of the separation effect. The national average refers to the positive rate of bacteria obtained through national surveillance. It also includes national averages stratified by hospital characteristics of the same size and type (emergency hospital, acute hospital, chronic hospital, geriatric hospital, pediatric hospital, etc.).
(2) Aggregation is performed in a plurality of predetermined observation widths, for example, observation widths every day, 1, 7, 14, and 30 days.
(3) Since the test results of bacteria take days to be finalized, for the sake of immediacy, recalculate every time an interim report is made.

More specifically, for example, as shown in FIG. 1, first, all the hospitals to be detected, each ward, each department such as each department (for example, hospital A, hospital B, hospital C..., Ward) A, Ward B, Ward C ..., Clinical Department A, Medical Department B, Medical Department C ... ) , a plurality of predetermined observation period widths (so-called 1 day, 7 days, 14 days, 30 days, etc.) _{window) W 1, W 2, W} 3 specimen submission number of patients in ... (that is, the number of patients subjected to bacteriological examination) and the N _{sample-patient,} for all of the bacteria _{x 1, x 2, x 3} ··· The number N _{positive-patient} of each positive for the bacterium is counted (steps S11 and S12). This aggregation is performed every day or at a frequency of a predetermined period. For example, N _{sample-patient} and N _{positive-patient} for each observation period are aggregated in the database while sequentially repeating the observation period of W ₁ = 1 day, W ₂ = 14 days, and W ₃ = 30 over one year. ,input.

On the other hand, for example, as shown in FIG. 2, the positive rate R _{positive of} all the bacteria, that is, the probability that the bacteria are detected by examining one patient, for a sufficiently long period such as one year. The number of patients who are positive for the bacteria is divided by the number of specimen-submitted patients and obtained in advance. More specifically, first, the data aggregation period is designated (step S21), and the number of specimen-submitted patients and the number of positively detected patients for each bacterial species are totalized and input (steps S23 and S24), and the bacteria positive rate R _positive is calculated. (Step S25).
At this time, in the present invention, duplicate elimination of the same patient is performed so as to be the total number of months and people (step S22). More specifically, for example, as illustrated in FIG. 3, it is determined whether or not the same bacterial species is detected from the same patient within a predetermined observation period (for example, within 30 days) (step S31). Without deduplication process when it is not detected (step S33), when it is detected determines whether there is a summary of involving antimicrobial susceptibility (step S32). If there is no tabulation, no duplicate elimination processing is performed (step S33), and if there is tabulation, duplicate bacteria with resistance determination are eliminated (step S34). Specifically, the number of common antibacterial drugs is set to 0 as an initial setting (step S35), and it is determined whether there is a difference in resistance for each antibacterial drug (step S37). If there is no difference, the number of common antibacterial drugs = the number of common antibacterial drugs + 1 (step S38), the determination process is looped (steps S36-S38), and if there is a difference in resistance, the number is counted as another bacterium. (Step S311). After all the antibacterial drugs have been determined (step S36), it is determined whether the number of common antibacterial drugs is the same as the number of test antibacterial drugs of any strain (step S39). If they are the same, they are treated as the same bacteria (step S312). If they are different, it is determined whether the number of common antibacterial drugs is a predetermined value (for example, 5) or more (step S310), and is not greater than the predetermined value. In this case, it is counted as another bacterium (step S311), and when it is a predetermined value or more, it is processed as the same bacterium (step S312). Then, the duplicate elimination process is executed for the same bacteria, that is, duplicate bacteria are deleted from the total data (step S313). Subsequently, it is determined whether or not the number of the test antibacterial agents for the bacteria tested earlier in time series is equal to or greater than the number of the test antibacterial drugs for the bacteria tested later in time series (step S314). If this is true, the previous test is left (step S315), and if it is not true, the sample detection date of the previous test is inherited as the date of determination of the predetermined observation period (step S316), and the later test is left (step S317). ).
FIG. 4 is an example of the bacteria positive rate R _positive calculated according to each process including the above-described deduplication.

Assuming that the isolation of the fungus is unbiased and controlled only by chance, the probability P _isolation of each fungus from each number of patients at each location is the positive rate of that fungus is R _positives, in some places, there bacteria, the probability P _isolation separated from a patient of a certain number of people, and their number, and test the patient number N _{sample-patient,} binomial from prevalence R _positives of the fungus The distribution can be used (step S13 (FIG. 1)). Further, by adding a probability that a number smaller than that number is separated, that is, using a cumulative binomial distribution, for example, it is possible to obtain a probability that up to 4 people are positive for the bacteria by examining 6 people. In the present invention, the probability is calculated every day for every place, every observation period, and every bacterial species.

The purpose of providing a plurality of observation period widths is that neither a short sharp change nor a long-term gentle change is overlooked. Small accumulations can be specifically detected by continuously calculating the probability every day with these widths.
For example, as shown in FIG. 5, a more specific calculation process of the fungal isolation probability P _isolation is set to P = 0 as an initial setting (step S51) and m = k to n (steps) as shown in FIG. S52) The following calculation is performed. First, _n C _m (= C _x ) is calculated (step S53), and the probability p ^m (1−p) ^nm (= P _tankai ) of exactly m bacteria in the al order is obtained (step S54). The probability C _x P _tankai (= P _kumiawase ) of individual bacteria is calculated (step S55), and P = P + P _tankai is set (step S56).

In the present invention, the calculated _isolation probability P _isolation is further classified according to its value, and the risk of abnormal accumulation (warning level), the location where the problem occurred, the date and time of counting, the observation period width, the bacterial species, the number of positive patients The number of patients who submit specimens, probability, etc. are displayed as warnings.
More specifically, threshold values (a plurality of levels) for the separation probability P _isolation are designated in advance (step S14 (FIG. 1)), and the bacteria separation events are classified by the separation probability P _isolation (step S15). Then, an event having an _isolation probability P _isolation below each level is warned as an abnormality (step S16).

  In addition, as an application of this technology, by combining with the antibiogram classification technology and detecting abnormal accumulation for each classification, this can be detected automatically when specific resistant bacteria spread in the hospital or in the area. It becomes possible.

[Antibiogram classification]
The principle of antibiogram classification using SIR determination in the present invention is as follows.

  As a result of examining the treatment of I (intermediate) with ambiguity in the SIR determination, I can take either S (sensitivity) or R (resistance), so it is not a determinant and is not being tested. It turns out that the same treatment is acceptable. It turned out that the method of scoring and judging that they are the same is not practical because the final combination has ambiguity. Therefore, the antibiogram classification of the present invention covers combinations that cannot be denied that logically (that is, rule based algorithm) belong to the same group.

More specifically, as shown in the flowcharts of FIGS. 6 and 7, for example, among the combinations of two bacteria A-B (step S 61 , steps S 71 -S 72), at least one difference between S and R is different. Certain combinations, that is, combinations that are not S or R to each other, can be ruled out as belonging to the same group (steps S62-S63, step S73), and there is no number of matching antibacterial drugs between these two bacteria The number of inspections is set to 0 (Step S64, Step S74). For the other cases, that is, for the combinations that are S or R for each other, the number N of antibacterial drugs that match between the two bacteria is obtained and set as the number of examinations N (steps S65, S75-S76). A certain standard (for example, 3 or more) is set for the number of tests that match, and a determination table is created by determining whether or not all the bacteria match. Based on the determination table, a group of bacteria whose number of tests that match each other exceeds a certain standard is created (steps S66-S67, steps S77-S78). To make a combination, a method using set theory is used.

  More specifically, for example, the following processing is performed.

  (1) Treat SIR determination I as no data input, and determine only whether S and R match or not.

(2) Compare all two combinations of patterns and eliminate if there is any difference in tolerance S, R (set to 0 in (3))
(3) The number of antibacterial drugs having the same resistance is scored, and the identity of two patterns having the number of antibacterial drugs equal to or greater than a threshold (for example, 3 points) is determined.

  (4) Three or more combinations are obtained from the identity that is the relationship between the two patterns.

  Here, for example, an example of a specific algorithm for obtaining three or more combinations from the identity which is the relationship between two patterns is as shown in FIG. The set Ai is a set of strains whose resistance pattern identity cannot be denied with the strain xi. When strain xj is an element of set Ai, strain xi is an element of set Aj. Given Ai and Aj, the third strain that cannot be denied that the resistance pattern is the same is a common set. Therefore, the combination of the third and subsequent strains can be obtained with high efficiency by having, in the main memory, a set of combinations whose score obtained by scoring the number of antibacterial drugs having the same resistance in the main memory.

For example, FIG. 9 and FIG. 10 show the result of the automatic classification of the antibiogram according to the present invention, FIG. 9 is a list of the number of matching test results, and FIG. 10 is the automatic classification result. It can be seen that the matching resistant drug score as shown in FIG. 10 is obtained based on the SIR total data for each detection target place and each specimen-submitted patient in FIG.
More specifically, in step S73 in FIG. 7, for example, when comparing the SIR of bacteria K41 and K70 in FIG. 10, K41 is R and K70 is S with respect to the drug AZT, so K41-K70 does not match. The combination is determined, and in step S64, K41-K70 is a mismatched combination and is set to 0 (see FIG. 9). On the other hand, in step S65, for example, when comparing the SIRs of the bacteria 41 and 42 in FIG. 10, first, the drugs AZT and CTX in which at least one of the bacteria is I are excluded, that is, K42 is I for the drug AZT. Yes, since both K41 and K42 are I with respect to the drug CTX, the S / R between K41 and K42 for AZT and CTX is excluded from the determination. Subsequently, for other drugs, the drugs AMK, GM , Since it is S for all of the PIPCs, it is determined that K41-K42 is a matching combination. In step S66, K41-K42 are the same for the three drugs AMK, GM, and PIPC, so that three points are set (see FIG. 9). Similarly, for example, K42-K50 are two points because they match in the two drugs AMK and GM (see FIG. 9).
And although the result of grouping is not shown in the figure, the grouping process as described above can realize the grouping of bacteria that cannot be denied that the resistance pattern is the same, that is, the antibiogram classification of the bacteria it can.

  In addition, as an application of the above antibiogram classification technology, practicality increases by combining with the following two-dimensional career map.

  In addition, it is conceivable to add a mechanism for pointing out temporal and spatial accumulation (pointing out the risk of spreading specific resistant bacteria) to grouping.

If the average resistance rate of each drug can be obtained, it is possible to weight the matching rate of the antibiogram. It is possible to add a mechanism to use as an index of probability.
By the way, with regard to the above automatic classification of antibiograms, the amount of processing can be dramatically reduced by pre-processing the data to be processed before the main processing illustrated in FIG. Specifically, as illustrated in FIG. 11, first, it is determined whether the result of the SIR determination is the same for all bacteria, that is, all R or I, or all S or I (step S111). -S113), for example, the corresponding antibacterial drug is excluded from the table of FIG. 8 and excluded from the classification processing target (step S114). Subsequently, bacteria with exactly the same sensitivity pattern are treated as one. That is, a group of bacteria having the same combination of SIRs is grouped and the representative bacteria are tabulated (step S115). Then, after the classification process as shown in FIG. 6 (or FIG. 7), the representative results grouped in step S115 are copied to the entire group obtained in steps S66- S67 and steps S77- S78 (step S116). In the preprocessing, when there are a plurality of antibacterial drugs having the same SR pattern for all the specimens, a process of leaving only one as a representative may be executed.
FIG. 12 shows an example of representatives grouped by this preprocessing, FIG. 13 shows an example of classification results when the grouped representatives of FIG. 12 are copied to the entire group in S66 and S77, and FIG. 14 shows the result of color coding. An example is shown.

[Create 2D carrier map]
The principle of creating a two-dimensional carrier map in the present invention is as follows, for example, as illustrated in FIG. FIG. 16 is an example of a two-dimensional carrier map created based on the classification color code of FIG. 14 according to the present principle.

  First, the antibacterial drug classification based on the antibiogram classification, that is, the group of bacteria that cannot be denied by the antibiogram classification is the color code. A different color may be set for each group.

  Next, the separation data of each bacterium is plotted along the time axis and the space axis on the map based on the three parameters of the separated patient, location, and time.

  Then, each plot of the same patient is connected with a line.

  More specifically, the following processing is performed along FIG.

  (1) After automatically classifying the antibiogram, a different color is selected for each group of bacteria, and the group and color are associated and stored (step S151).

  (2) A two-dimensional carrier map is prepared in which the X axis is a time axis and the Y axis is a list-up axis for all specimens. Along the Y axis, all samples are sorted based on patient information, location information (buildings, floors, wards, hospital rooms, etc.) prerecorded for each sample, and sample collection date and time as a list on the left side of the map They are displayed vertically arranged in one column (step S152). Of course, XY may be a reverse combination.

  (3) Plotting at the height of the Y position of the list at the X position corresponding to the sample collection date and time (step S153). Each plot is colored with an antibiogram classification grouping (step S154). If you belong to multiple antibiogram classification groups, arrange them horizontally or vertically as needed.

  (4) Linear data is assigned between plots of specimens from the same patient and displayed on the map (step S155).

  As described above, it is possible to analyze the state in which the bacteria having the same resistance pattern are diffused in the ward or the like and grasp it very easily.

  For example, in the example shown in FIG. 16, the bacteria classified into brown and olive in the round frame are hardly detected from other than this patient, so the same from another patient admitted to the same ward at the same time. If the bacteria of the classification comes out, it turns out that nosocomial infection cannot be denied.

[Evaluation of infection control index by PID method]
The principle of the PID infection countermeasure index evaluation in the present invention is as follows.

  As the original data of PID control, numerical values that are indexed (standardized) for infection countermeasures, such as the isolation rate of bacteria, the number of positive patients, the number of patients with infectious diseases, the number of patients with positive symptoms of infection such as fever, the number of patients with positive inflammation tests, The number of infectious disease test positive patients and their rate, for example the incidence of infectious disease, are used.

  More specifically, for example, as shown in FIG. 17, first, target values are set for various indicators that can be graphed as original data for PID control (step S171).

  Next, the deviation from the target value, the accumulation of deviation from the target value (time integration), and the speed of change (time differentiation) are calculated. Specifically, for each predetermined observation period unit, a P factor indicating a difference (deviation) from the target value, and a D factor indicating a difference from the previous observation unit on the time axis (speed of change, time differentiation) ), An I factor (accumulation of deviation, time integration) indicating the sum of deviations in a predetermined period in the past is calculated (step S172).

  Subsequently, the sum of the calculated P factor, D factor, and I factor is set as a PID value (step S173). Here, it is preferable that the period for taking the sum of the factor I is an arbitrary past period or a predetermined observation period width.

Then, according to the PID value, that is, the PID control is executed using the PID value as an index of the degree of effort to make the infection countermeasure result the target value (step S174), the PID control always uses the index. It is required to move in a good direction (small or large). When the target is in the direction of decreasing values, it is not necessary to increase the value ("make infection control worse") just because the measured value has become lower than the target. Therefore, it is necessary to use a “PID positive value” that takes the absolute value of each value.
PID control is a set of standardized observation values such as time axis, observation value, and rate. For data that can be graphed, a target value is set and the change in data over time is determined based on the deviation from the target value. The purpose is to ascertain the trend with certainty. A method of obtaining a deviation (P factor), an integral value (I factor), and a differential value (D factor) for an observation value taking values x ₀ to x _n from time t ₀ to t _n will be described. _{Incidentally, P k, I k, D} k, PID k denotes P, I, the values and PID values of Factor D at time _{t k, Pa k, Ia k} , Da k, PIDa k is equivalent P, I, D positive value and PID positive value are shown. An example of a specific mathematical formula is as follows, and each processing step according to this formula is as shown in FIG.

  In the present invention, the integration period can be specified.

[Graphization of cumulative detection warning score of bacteria automatic accumulation]
The principle of graphing the accumulated detection warning score of the bacteria automatic accumulation in the present invention is as follows.

  Using the above-described detection of abnormal accumulation of bacteria using the binomial distribution in the present invention, the warning score obtained as the output is accumulated and accumulated. The accumulation of scores is totaled for each fixed period (for example, every month) using the score of warning for accumulation as an index, and graphed.

Specifically, first, the warning score obtained as an output by the abnormal accumulation detection of the bacteria using the binomial distribution in the present invention described above is totaled for each bacteria for a certain period (for example, one month). Here, the warning score is obtained by classifying the probabilities calculated based on the null hypothesis that the bacterial separation is sporadic with a threshold, and giving a high score to a low probability. For example, p <0.001: 3 points, p <0.005: 2 points, p <0.01: 1 points.
More specifically, for example, as illustrated in FIG. 19, first, a period for plotting is specified (step S191), and the handling of bacteria having the number 1 of bacteria-positive patients is specified, that is, not counted (step). S192). The bacteria to be plotted are designated (step S193), and a monthly total (warning score accumulation) table of the designated period, the designated counting method, and the warning score of the designated bacteria is created using the warning score table (step S194). A warning score table including the counting date, bacteria, the number of bacteria positive patients, the number of patients to be tested, and a warning score is stored in advance in a computer system such as an infectious disease monitoring server (database or the like). The generated warning score accumulation table is also stored in the computer system.

Then, for example, a graph as illustrated in FIG. 20 is created based on the result of the aggregation, that is, the warning accumulation table data (step S195). In the graph of FIG. 20, the horizontal axis represents the time axis, and the vertical axis represents the warning score. The graph shown in FIG. 20 shows an example of simulation related to Serratia marcescens, and it is confirmed that Serratia marcescens has accumulated abnormally in the hospital almost every six months over six years. I understand.
Here, by identifying the bacterial species that repeat accumulation from this warning score accumulation graph, and combining epidemiological characteristics (for example, well-isolated materials, infection routes, etc.) of those bacteria, infection countermeasures to be focused on Can be specifically and automatically indicated.
First, as illustrated in FIG. 21, a sum of warnings (levels) of automatic detection of abnormal accumulation of bacteria every month is set as a cumulative warning score.
Next, as illustrated in FIG. 20, the warning score accumulation is graphed. Thereby, it is possible to overlook the diffusion situation of bacteria in the hospital for a long time.
And if the warning score accumulation is totaled about all the isolation microbes, the microbe which repeats in-hospital spreading | diffusion can be known. The bacteria are different from each other depending on what kind of material is well separated. For example, Escherichia coli is largely separated from stool and urine, and staphylococci are largely separated from the skin and nasal cavity. About the bacteria that have repeatedly spread in the hospital, from what kind of material is it isolated, and knowing the material from the separation frequency for each test material for each bacteria, know the source of the infection causing the spread be able to. In the counting example of FIG. 22, in this facility, in recent years, Citrobacter, Enterococcus, and Enterobacter have been repeatedly diffused in the hospital. Expected to be. Since the separation frequency for each bacteria and each test material can be easily obtained from the data of the hospital or the national data, it is possible to automatically determine the common test material as an infection source. In addition, some bacteria have well known diffusion routes, and it is possible to improve the accuracy of such bacteria by combining them with a knowledge database in which the bacteria and main infection routes are input.
According to this, the infectious spread of these bacteria by strengthening infection control measures regarding the infection source as the infectious source that the bacteria are spreading in the hospital, using the test material from which many bacterial species that have repeatedly spread in the hospital are isolated It is possible to prevent in-hospital infection accidents (outbreaks).
In other words, analysis of bacteria isolated from hospitals has been repeated in the hospital and analyzed for “problem bacteria” that are likely to cause nosocomial infections. Can show.
Since bacteria can hardly move by themselves, some kind of human intervention is required for the bacteria to reach the entrance (establishment) gate of the infected person as the first step of infection.
A six-element model of infection (Jackson MM. General principles of epidemiology.; APIC infection control and applied epidemiology: principles of practice. St. Louis: Mosby, 1996; 1: 1: 1-19) The steps to enter the patient are classified in an easy-to-understand manner. The source of infection, the route of excretion, the route of infection, the route of entry, and the person to be infected vary depending on the pathogen (fungus), for example, one bacterium is resident in the stool, another bacterium is resident in the nasal cavity, and Even if it does not exist permanently, the material to be separated is determined mainly by the difference in pathogenicity. On the other hand, pathogens, infection sources, excretion routes, infection routes, invasion routes, and infected persons have some relationship to each other. For example, even if the excretion route is the same respiratory organ, the infection route is due to droplet infection. In some cases, in order to move with air by finer aerosols, each step is independent to some extent, such as when an infection route called air infection is taken or contact infection is taken.
Infection control is to prevent infection by intervening in pathogens, infection sources, excretion routes, infection routes, invasion routes, and links of infected people.
Therefore, the most important point is to find out which factors and what types of factors are weak points for infection control by finding common factors for bacteria that have been repeatedly spread in the hospital. Is to do.
Therefore, the detection method of the bacteria that repeats in-hospital diffusion and the algorithm for finding the common term of the above factors are not particularly limited, but an example thereof will be described below. In addition, it is considered useful to add elements such as infection frequency in addition to the six elements.
FIG. 23 is a diagram for explaining scoring of bacteria that repeat in-hospital diffusion.
<Step S221>
First, as an initial process, the target bacterial list is emptied. The problem bacteria are those that repeat in-hospital diffusion such as the cumulative total score of the warning score, and the target bacterial table is a table of these, the diffusion obtained from the bacterial name (bacteria code) and the cumulative total score of the warning score, etc. A numerical value indicating the risk level (risk rank: a higher risk level is set to a higher value). There is no limitation on how to obtain the danger rank. The target bacteria table is stored in advance in a computer system such as an infectious disease monitoring server (database or the like).
<Step S222>
From the warning score cumulative total score table (see FIG. 21), high-scoring bacteria are selected and added to the problem bacteria table. A high-scoring bacterium refers to a bacterium having a large warning score cumulative total score. For example, from the top, 10 bacterial species are designated as high score bacteria. Alternatively, for example, bacteria having a large numerical value can be selected by a plurality of methods, for example, a bacteria having an average of 20 or more per month is selected as a high score bacteria.
<Step S223>
From the warning score cumulative total score table (refer to FIG. 21), a score increasing tendency bacteria is selected and added to the problem bacteria table. The bacteria with a tendency to increase score are not high score bacteria in the total of the warning score cumulative total score, but for example, the score is high when limited to the last few months, and the warning score cumulative total score is higher than the past It refers to bacteria that tend to increase.
<Step S224>
Create a table of the target bacteria and their infectious agents (excretion route, infection source, infection route, intrusion gate, infected person, etc.). Specifically, it is created by combining the target bacteria table created by adding each bacterium as described above and the bacteria / infectious agent table recorded in advance. The known bacteria / infectious agent table is used, and the infectious agent of the corresponding problem bacteria is searched from the fungus group in the list, and added to the problem bacteria / infectious agent table.
<Step S225>
The created problem bacteria / infectious agent table is read from the database and displayed on a computer display or the like.
<Step S226>
Create a table of scores by infectious agent and content (number of bacteria or total risk rank of the bacteria). The specific creation process is as follows.
(1) First, each column of each table of infectious agents (excretion route, infection source, infection route, intrusion gate, infected person, etc.) is given 0 points.
(2) First, the first target bacterium in the table of target bacterium / infectious agents obtained in step 224 is the target of processing.
(3) For each of the first target bacteria, add the danger rank of the bacteria to the column of the nature of the bacteria in each score table. For example, if contact infection is to be taken for the infection route, add 1 to the contact infection in the score table of the infection route for the risk rank (for example, 12) of the fungus or the count that does not use the risk rank. If so, the same rank in the mouth column of the score table for the intrusion gate (add the same value in any table you add, 12 or 1 in the previous example).
(4) Next, the second target bacteria are processed in the same manner, and sequentially processed until all the target bacteria are completed.
<Step S227>
Sort the table of scores by infection factor and content (number of bacteria or total risk rank of bacteria) in descending order.
<Step S228>
The infectious agent-specific and high-scoring elements are read from the database and displayed on a computer display or the like. Here, it is preferable to color-code by score and score category. The specific display process is as follows.
(1) First, the first factor of the infectious factor (excretion route, infection source, infection route, intrusion gate, infected person, etc.), for example, the excretion route is the target of processing.
(2) Since the score is obtained for each excretion route such as urine, feces, milk, blood, saliva, cough, sputum, and skin in the score table, the table is rearranged in descending order of score. Display a table with a high-priority color such as red.
(3) Next, other factors such as an infection source are targeted for processing, and the same processing is performed. Sequentially, processing is performed until all the target bacteria are finished.

  As described above, for example, by grasping the accumulation status from several months to several tens of years, (1) problem discovery and countermeasure planning, and (2) countermeasure effectiveness determination become possible, which is useful for hospital infection countermeasures. (3) By grasping the status of abnormal accumulation of all isolated bacteria over several months to several decades, it is possible to grasp the problem of countermeasures from the main infection modes of multiple bacterial species that have been identified as abnormal accumulation only. It becomes possible, and effective measures can be planned and useful. Similar aggregation is possible in the national surveillance system, and the same effect can be obtained.

[Device configuration]
Each process in the present invention as described above can be executed by an apparatus having the overall configuration illustrated in FIG. 24 includes a processing unit (CPU) 10, a storage unit (internal memory, external memory) 11, an input unit (keyboard, mouse) 12, an output unit (monitor, printer) 13, and a communication control unit 14. Is connected. The device 11 is connected to the network 16 via the communication control unit 14. It is possible to obtain various data necessary for each process stored in another terminal or the like from a terminal such as the device 1 connected to the network 16. The processing unit 10 executes the various processes described above by reading the program stored in the storage unit 11.

The flow chart of abnormal bacteria accumulation detection for explaining the present invention. The flowchart of a microbe positive rate calculation process. 10 is a flowchart of deduplication processing. An example of the calculation result of a bacteria positive rate. The flowchart of a microbe isolation | separation probability calculation process. The flowchart of the automatic classification of antibiogram for demonstrating this invention. A more specific flowchart of automatic antibiogram classification. The conceptual diagram for demonstrating an example of microbe grouping in an antibiogram automatic classification | category flow. The result of automatic classification of antibiogram according to the present invention. The result of automatic classification of antibiogram according to the present invention. The flowchart of the pre-processing and post-processing in antibiogram automatic classification. An example of a representative grouped by preprocessing. An example of a classification result. An example of color coding. The flowchart of preparation of the two-dimensional carrier map for demonstrating this invention. An example of the two-dimensional carrier map by this invention. The flowchart of the infection countermeasure parameter | index evaluation by PID method explaining this invention. The flowchart of a P, I, D positive value, and PID positive value calculation process. The flowchart of the graphing of warning score accumulation of the bacteria abnormal accumulation detection for demonstrating this invention. An example of the graph of the warning score accumulation | accumulation of bacteria abnormal accumulation detection by this invention. An example of warning score accumulation. Another example of a warning score accumulation graph. The flowchart of scoring of the microbe which repeats the in-hospital spreading | diffusion by this invention. The system block diagram for demonstrating this invention.

Claims (10)

By the processing unit of the system including the processing unit, the storage unit, the input unit, the output unit, and the communication control unit,
For each place to be detected for abnormal accumulation of bacteria, the bacteria were positive for each bacteria for the specimen submitting patient and all the bacteria to be detected during the predetermined observation period at the specified frequency. Total data calculated in advance by counting the number of patients,
For all bacteria, using a binomial distribution based on the total number of positive patients for each bacterium, the number of patients submitting specimens, and the baseline rate , which is the positive rate for each bacterium determined in advance, The probability of the successful Bernoulli trial is calculated according to the null hypothesis that there is no bias in bacterial isolation and that bacterial isolation is controlled only by chance. It is judged that there is an abnormal accumulation with a small probability of being ruled and biased separation of bacteria,
A method for detecting abnormal accumulation of bacteria,
In the processing unit, when the baseline rate is calculated, deduplication of the same patient is performed so as to be the total number of months and people, whereby the baseline rate is a single time of the binomial distribution. To be the probability of independent trials,
A method for detecting abnormal accumulation of bacteria.   The method for detecting abnormal accumulation of bacteria according to claim 1, wherein a positive rate of the bacteria at the place or a national average positive rate is taken as a baseline rate.   The method for detecting abnormal accumulation of bacteria according to claim 1, wherein the aggregation is performed at a frequency of consecutive days or a predetermined period with a plurality of predetermined observation widths. For each place to be detected for abnormal accumulation of bacteria, the bacteria were positive for each bacteria for the specimen submitting patient and all the bacteria to be detected during the predetermined observation period at the specified frequency. Total data calculated in advance by counting the number of patients,
For all bacteria, using a binomial distribution based on the total number of positive patients for each bacterium, the number of patients submitting specimens, and the baseline rate , which is the positive rate for each bacterium determined in advance, The probability of Bernoulli trials to be established is calculated according to the null hypothesis that there is no bias in bacterial isolation and that bacterial isolation is controlled only by chance,
When the calculated probability is smaller than the default value, the probability of being controlled only by chance is small, and it is determined that there is an abnormal accumulation due to bias in the separation of bacteria,
A device for detecting abnormal accumulation of bacteria,
When calculating the baseline rate, by deduplicating the same patient so as to be the total number of months and people, the baseline rate is determined as the probability of a single independent trial of the binomial distribution. To be,
An apparatus for detecting abnormal accumulation of bacteria characterized by the above.   The apparatus for detecting abnormal accumulation of bacteria according to claim 4, wherein a positive rate of the bacteria at the place or a national average positive rate is taken as a baseline rate.   The abnormal accumulation detection apparatus for bacteria according to claim 4, wherein the aggregation is performed at a frequency of consecutive days or a predetermined period with a plurality of predetermined observation widths. A method of accumulating and graphing a warning score obtained as an output of the method for detecting abnormal accumulation of bacteria according to any one of claims 1 to 3 , comprising:
Classifying the calculated probability of Bernoulli trials that represent a non-biased separation of the bacteria by a threshold, and summing the warning score obtained by giving a high score to those with a probability lower than the threshold, for each bacterium for a certain period, Graph the aggregated results with the horizontal axis as the time axis and the vertical axis as the warning score.
A graphing method for accumulating warning scores for detecting abnormal accumulation of bacteria. An apparatus for accumulating and graphing a warning score obtained as an output of the abnormal accumulation detection apparatus for bacteria according to any one of claims 4 to 6 ,
Classifying the calculated probability of Bernoulli trials representing a non-biased separation of the bacteria by a threshold, and a warning score obtained by giving a high score to a low probability, a means for aggregating the bacteria every fixed period, Means for graphing the aggregated results with the horizontal axis as a time axis and the vertical axis as a warning score;
A graphing apparatus for accumulating warning scores for detecting abnormal accumulation of bacteria, comprising:   A program for causing a computer to execute the method for detecting abnormal accumulation of bacteria according to any one of claims 1 to 3.   A computer-readable recording medium on which the program according to claim 9 is recorded.