GB2352815A

GB2352815A - Automatic health or care risk assessment

Info

Publication number: GB2352815A
Application number: GB9910052A
Authority: GB
Inventors: Keith Henderson Cameron
Original assignee: Individual
Current assignee: Individual
Priority date: 1999-05-01
Filing date: 1999-05-01
Publication date: 2001-02-07
Also published as: GB9910052D0

Abstract

An automatic risk assessment system for care, which is comprised of a sensor, or a set of sensors (A,B,C,D,E,F,X), which measure a person's physiological and psychological characteristics, which can be combined with, or used independently from, other inputs (Q,I,V,W) giving further care related data. For example, sensor (B) measures walking velocity and (I) gives a measure of the Activities of Daily Living assessment. Data from these sources are used to derive care indices, such as mobility (R). Data from these indices, together with data from other sources are used in the predictive risk processor (S), to derive the risk of disease, illness, or accident as a predictive index based on an adaptive algorithm, itself derived from a weighted combination of the known risks. These being sourced from learned sources and from feedback from the system. The value of the predictive risk index (Y) is then used to provide the most appropriate level of intervention for the person at risk. Although the invention is intended to predict care problems due to falls associated with the elderly, it may apply to the prediction of other illnesses or diseases, e.g. heart or cardio disease.

Description

2352815 PREDICTIVE RISK SYSTEMS FOR CARE The present invention relates to

methods and arrangements for an automatic risk assessment system, which in particular but not exclusively can predict, present or future risk as a result of interpreting a set of measured, pre-detennined or pre-existing factors or derived care indices, and then can summon appropriate intervention for a person who as a result of this assessment is determined to be at risk.

Many authors, such as Besdine, Williamson and Isaacs [see, Besdine RW, Geriatric Medicine, "An Overview; Annual Review of Gerontology and Geriatrics"; 1, 135153, 1980; Williamson J et al, "Old People at Home: their Unreported Needs"; Lancet, 1, 1117, 1964; Isaacs B, Gunn J, McKechan A, McMillan I, Neville Y, "The Concept of Pre-death", Lancet, 3, 1115, 1971] have indicated that the present system of health care based on the principle of self-reporting is especially vulnerable when it attempts to meet the health needs of an ageing population whilst preventing institutionalisation. Hence, the inadequate reporting of illness by the elderly makes it much more probable that disease will be far advanced before the elderly patient gets into the health care system. During this delay between onset and detection, it is possible for multiple pathologies to interact, harming the patient and producing irrevocable disability in spite of eventual excellent care.

The present invention relates to a system which if added to the current self-reporting health care systen-4 could provide active case-finding surveillance mechanisms. The present invention relates to a system which collects, collates, and interprets data from a sensor or sensors, from other data sets, and from derived care indices, which enables a measure of present or future risk to be determined. This measure can then be used to alert or inform, at the most appropriate level; either, or a combination of, the person at risk, their caters, care equipment or systems; or other devices, people or systems which could make use of such information. By way of example, data from a variety of measurement devices or sensors, other data sets, and derived care indices relating, by way of example, to physiological, psychological or other care related parameters, could be used to derive a predictive index of the likelihood of a person being at risk of disease, illness or accident in the future.

By way of example, parameters identified by Isaacs [referenced above], that is incontinence, immobility and cognitive impairment, could be used as care indices to derive predictive indices forewarning of the on-set of future possible illness or disease. Also, by way of example, monitoring parameters which predispose people to a subsequent fall could be used to derive a predictive fall index.

According to the present invention there is provided a predictive risk system for care. This comprises a single sensor, or a set of sensors, which could be used separate from, or combined with, additional data from other data sets, such as, by way of example, a record of personal data, environmental data, test data or medical infon-nation. Some or all of this data can be used as inputs to derive care indices, which could be, by way of example, a mobility index. The care indices are determined by processing the data inputs in combination or separately to derive a representative care status for the particular index. By way of example, this could be carried out by an operator or by using simple logic or artificial intelligence techniques. Alternatively or additionally, the data, together with data from the care indices, where appropriate, can be used as inputs for an algorithm or other decision processing methodologies, the output from which gives a prediction of the likelihood of risk. The data inputs can be at various data rates, by way of example, in real time, time delayed, time averaged, yearly, or fixed. The care indices and the predictive index outputs can also be of various forms and data rates. By way of example, a rolling average of an index on a daily basis, or a measured index on a weekly basis. By way of example, the prediction algorithm or other decision processing methodologies could be based on a weighted combination of input data. The weighting factors, by way of example, being determined by a "best understanding" of the risks involved in predicting future disease, illness or accident. This " best understanding" itself being derived from a number of factors. By way of example, the pre-disposition to falling has been studied extensively in the literature [see by way of example, Teno J, Kiel D, Mor V "Multiple Stumbles: A Risk Factor for Falls in Community-Dwelling Elderly" Journal of American Geriatrics Society, 38, 12, 1321-1325, 1990; Koski K, Luukinen H, Laippala P, Kivela S "Physiological Factors and Medications as Predictors of Injurious Falls by Elderly People: A Prospective Population Based Study" Age and Ageing, 1996, 25, 29-3 8; Hombrook M, Stevens V, Wingfield D, Hollis J, Greenlick M, Ory M "Preventing Falls among Conununity Dwelling Older Persons" The

Gerentologist, 1994, 34(l), 16-23; Tinetti M, Mendes de Leon C, Doucette J, Baker D "Fear of Falling-Related Efficacy in Relationship to Functionality among Conununity Living Elders" Journal of Gerontology: Medical Sciences, 1994, 49(3), M140-MI47; Tinetti M, Doucette J, Clause E "The Contribution of Predisposing and Situational Risk-Factors to Serious Fall Injuries" Journal of the American Geriatrics Society, 1995, 43(11), 1207-1213; Studenski S, Duncan P, Chandler J, Samsa G, Prescott B, Hogue C, Bearon L "Predicting Falls: The Role of Mobility and Nonphysical Factors" Journal of the American Geriatrics Society, 1994, 42(3) 297-302; Asada T, Kariya T, Kinoshita T, Asaka A, Morikawa. S, Yoshioka M, Kakuma T "Predictors of Fall- Related Injuries among Conununity Dwelling Elderly People with Dementia" Age and Ageing, 1996, 25, 22-28; Lauritzen J, McNair P, Lund B "Risk Factors for Hip Fractures-A Review" Danish Medical Bullet4 1993, 40(4), 479-485] and these, and other studies, have determined quantitatively those factors which generate the highest risk of a subsequent fall.

A specific embodiment of the invention, by way of example, a predictive system for the risk of a fall, will now be described, with reference to the accompanying drawing in which:

Figure I shows a block diagram of the fall prediction system Referring to the drawing, Figure I shows a block diagram of a predictive system for the risk of a fall, in which A,B,C,D,E,F represent a particular group of sensors which measure respectively; incidence of impacts or knocks and/or falls, walk-velocity, bed use, chair and body mass, movement and drawer use. The outputs from these sensors being K,L,M, N,O,P which respectively provide data on; number of impacts or knocks and/or falls, timed walks, time spent in bed, time spent in a chair and changes in body mass, number of visits to a room, and the number of times a drawer is opened. Q represents data derived from cognitive tests or sensors designed to measure these parameters. Data streams L,M,N are used as inputs for R, which is the processing element designed to provide the mobility index data T. Similarly, data streams 0,P,Q are used as inputs for J, which is the processing element designed to provide the cognition index data U. Personal data or information input to G,H, by way of example, manually or via a computer link, gives data on age V and gender W, 4- respectively. Data on the Activities of Daily Living (ADL) score 1, can be input, by way of example, manually or via a computer link, or could be automatically determined using a sensor set represented by X, some of this sensor set may have common components in the sensors A,B,C,D,E,F. All of this data is used to provide inputs to S, which is a processor which uses this data, together with other data, by way of example, a real time clock, to provide a fall prediction index (FPI) derived from a fall risk algorithm. The fall prediction index can then be sent via a data output device Y, which could be, by way of example, an automated telephone message, to the most appropriate carer or carers. The initial fall risk algorithm is derived from averaging of the coefficients of the adjusted odds ratios for key risk factors identified in the literature, and from other learned sources. These averages are then used to calculate appropriate weights for each risk factor. This algorithm is dynamic in nature enabling it to be revised from up-dated data in the literature, from other learned sources and by feedback data from the systems own predictions, interventions and outcomes. By way of example, Table 1 shows the detail of some of the most significant risk factors, for falls in females, together with their calculated coefficients. The data is taken from the studies from the learned papers referenced above.

Age Gender Mobility ADL Cognition Falls 1.36 2.56 0.87 1.40 1.20 1.20 1.00 1.00 1.30 3.90 1.60 2.00 1.80 1.80 2.20 1.40 1.34 1.50 1.45 1.00 1.75 3.23 Average values 0.13 0.15 0.14 0.10 0.17 0.31 initiai weights Table 1 Female Risk Factors for Falls The initial female fall risk algorithm is therefore given by:

FPlfemale= 0.13(la) +0.15(,g)+0.14(Im)+O.I(Iadl)+0.18(lc)+0.33(ID Where: Ia is the age index Ig is the gender index Im is the mobility index ladl is the index derived form the Activities of Daily Living Ic is the cognition index If is the previous fall index

Claims

1. A predictive risk system for care, which comprises: a means for sensing, measuring or inputting fixed and/or varying physiological, psychological or other care parameters, a means for measuring or deriving care indices, a means of interpreting a set of measured, pre-determined or pre-existing factors, a means for deriving an initial risk assessment algorithm, a means to process the algorithm to derive the risk factor, a means of interpreting this risk factor in the form of a predictive risk index, a means of determining a response to that index, a means to alert or inform, at the most appropriate level either, or a combination of, the person at risk, their carers, care equipment or systems, or other devices, people or systems which could make use of such information, a means for modifying the risk assessment algorithm using revised data or following intervention or by feedback from the risk assessment system itself.

2. A predictive risk system for care, as claimed in Claim 1, where the means for sensing, measuring or inputting fixed and/or varying physiological, psychological or other care parameters, can comprise of a sensor or a combination of sensors.

3. A predictive risk system for care, as claimed in Claim 1, where the means for sensing, measuring or inputting fixed and/or varying physiological, psychological or other care parameters, can comprise of data entered manually or from a computer or other information handling devices.

4. A predictive risk system for care, as claimed in Claim 2 and 3, where the data inputs can be at various data rates or formats.

5. A predictive risk system for care, as claimed in Claim 2, 3 and 4, where the data inputs can be real time, time delayed, time averaged, yearly, or fixed.

6. A predictive risk system for care, as claimed in Claim 1, where the means for measuring or deriving care indices is derived from processing data, as claimed in Claims 2 and 3.

7. A predictive risk system for care, as claimed in Claim 6, where the data outputs of the care indices can be at various data rates or formats.

8. A predictive risk system for care, as claimed in Claim 7, where the data outputs of the care indices can be real time, time delayed, time averaged, yearly, or fixed.

9. A predictive risk system for care, as claimed in Claim 6, in which the processing of the data to derive the care indices can be carried out using human interpretation, logic, artificial intelligence, neural networks or other decision processing systems, or a combination of these methods.

10. A predictive risk system for care, as claimed in Claims 1,2,3,4,and 5 where the means of interpreting a set of measured, pre-determined or preexisting factors, is an algorithm or other decision processing methodologies.

11. A predictive risk system for care, as claimed in Claim I and 10, where the means for deriving an initial risk assessment algorithm is based on the use of a weighted combination of input data.

12. A predictive risk system for care, as claimed in Claim 11, where the initial weighting factors are determined by a "best understanding" of the risks involved in predicting fature disease, illness or accident.

13. A predictive risk system for care, as claimed in Claim 12, where the "best understanding" of the risks is derived from statistical values determined from recognised learned journals or other sources of expert knowledge, or from data derived from predictive systems.

14. A predictive risk system for care, as claimed in Claim 13, where the statistical values of a given risk are given by a mean, median or modal value or from other derived statistical values, or from a combination of these.

15. A predictive risk system for care, as claimed in Claim 12, where the weighting factors are determined by a normalising factor.

16. A predictive risk system for care, as claimed in Claim 1 and 10, where the means to process the algorithm to derive the risk factor, can be carried out using human interpretation, logic, artificial intelligence, neural networks or other decision processing systems.

17. A predictive risk system for care, as claimed in Claim 1, where the means of interpreting the risk factor is indicated by a predictive risk index.

18. A predictive risk system for care, as claimed in Claim I and 17, where the means of determining an appropriate response to the predictive risk index is determined, entirely or in part, by its value.

19. A predictive risk system for care, as claimed in Claim 1 and 18, where the value of the predictive risk index is used to alert or inforrn either, or a combination of, the person at risk, their carers, care equipment or systems; or other devices, people or systems which could make use of such information.

20. A predictive risk system for care, as claimed in Claim 18 and 19, where the means to alert or inform, at the most appropriate level can be via a communication device.

2 1. A predictive risk system for care, as claimed in Claim 20, where the communication device or system can be telephonic, computer based, radio, acoustic, or be visibly displayed.

22. A predictive risk system for care, as claimed in Claim 1, 10 where the risk assessment algorithm can be modified from the initial risk assessment algorithm claimed in Claim 11.

23. A predictive risk system for care, as claimed in Claim 22, where the risk assessment algorithm can be modified by revised or new data from learned literature, from other learned sources and by data from the systems own predictions, interventions and outcomes.