CN115798685B - Depression diet management method based on food image segmentation - Google Patents
Depression diet management method based on food image segmentation Download PDFInfo
- Publication number
- CN115798685B CN115798685B CN202211628844.1A CN202211628844A CN115798685B CN 115798685 B CN115798685 B CN 115798685B CN 202211628844 A CN202211628844 A CN 202211628844A CN 115798685 B CN115798685 B CN 115798685B
- Authority
- CN
- China
- Prior art keywords
- food
- depression
- meal
- index
- foods
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 235000013305 food Nutrition 0.000 title claims abstract description 120
- 238000007726 management method Methods 0.000 title claims abstract description 26
- 235000005911 diet Nutrition 0.000 title claims abstract description 24
- 230000037213 diet Effects 0.000 title claims abstract description 24
- 238000003709 image segmentation Methods 0.000 title claims abstract description 12
- 235000012054 meals Nutrition 0.000 claims abstract description 51
- 230000036541 health Effects 0.000 claims abstract description 37
- 238000004458 analytical method Methods 0.000 claims abstract description 21
- 230000001430 anti-depressive effect Effects 0.000 claims abstract description 13
- 239000000935 antidepressant agent Substances 0.000 claims abstract description 13
- 229940005513 antidepressants Drugs 0.000 claims abstract description 13
- 230000011218 segmentation Effects 0.000 claims abstract description 13
- 238000000034 method Methods 0.000 claims description 22
- 239000011159 matrix material Substances 0.000 claims description 18
- 235000015219 food category Nutrition 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 9
- 230000000994 depressogenic effect Effects 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000007621 cluster analysis Methods 0.000 claims description 5
- 235000005686 eating Nutrition 0.000 claims description 5
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 2
- 208000020401 Depressive disease Diseases 0.000 claims 1
- 238000011156 evaluation Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 4
- 239000006185 dispersion Substances 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000012163 sequencing technique Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 230000036651 mood Effects 0.000 description 2
- 235000016709 nutrition Nutrition 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 208000019022 Mood disease Diseases 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000001684 chronic effect Effects 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000005118 dietary health Nutrition 0.000 description 1
- 230000035622 drinking Effects 0.000 description 1
- 235000006694 eating habits Nutrition 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention belongs to the technical field of digital health management, and particularly relates to a depression diet management method based on food image segmentation, which comprises the following steps: constructing a digital health management database based on depression meal indexes; acquiring physical health condition data of a current user; according to the physical health condition data of the current user, the data intelligent analysis is carried out by combining the health management database of the depression meal index, and the meal recommendation result of the current user is inferred; acquiring and storing physical health condition data and food images of a user through a database; performing intelligent segmentation on the image, and performing antidepressant index analysis on the segmented food by adopting a Fisher discriminant analysis method; clustering the screened foods, and setting a meal recommendation result according to the physical health condition information of the user and the clustered food types; according to the Fisher discriminant analysis method, the depression diet indexes of the foods are analyzed and screened, so that the food diet recommendation result is more accurate.
Description
Technical Field
The invention belongs to the technical field of digital health management, and particularly relates to a depression diet management method based on food image segmentation.
Background
Depression is a common mood disorder that can be caused by a variety of causes, with significant and persistent mood drops being the primary clinical feature, and mood drops being disproportionate to their situational context. Most cases of depression have a tendency to recur, and most of each episode can be alleviated, and some may have residual symptoms or be converted to chronic. There are studies currently showing that the causative factors of depression are also related to the patient's eating habits. With the improvement of the living material level of people, the dining table is often free from lack of food causing depression. Especially, a large amount of food is often provided on the dining table every year, and the drinking culture is maintained, so that the occurrence of depression is further promoted. Thus, how to manage the dietary health of patients is beneficial to the treatment of depressed patients.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a depression diet management method based on food image segmentation, which comprises the following steps: constructing a digital health management database based on depression meal indexes; acquiring physical health condition data of a current user; according to the physical health condition data of the current user, combining a digital health management database based on depression meal indexes, performing intelligent data query and reasoning to obtain a meal recommendation result of the current user;
the process of constructing a digital health management database based on depression meal indices includes:
s1: acquiring physical health condition data information and food pictures of a user;
S2: dividing the food pictures to obtain each food picture;
S3: performing depression diet index analysis on the separated food images by using a Fisher discriminant analysis method, and screening the number of foods with depression diet indexes;
S4: clustering the screened foods;
s5: setting a meal recommendation result according to the physical health condition information of the user and the clustered food category, and storing the physical health condition data information of the user and the meal recommendation result into a database.
Preferably, the process of performing the segmentation processing on the food picture includes: preprocessing the food image, wherein the preprocessing comprises filtering, enhancing and complementing the image; converting the preprocessed picture into a matrix; and performing edge segmentation on the matrix by adopting a Robert operator and a Sobel operator, and outputting an image and the matrix after the edge segmentation.
Preferably, the Fisher discriminant analysis method is adopted to analyze the depression diet index of the segmented food image, and the process comprises the following steps: all foods were divided into two sample subsets, two subsets being: a subset of food category data effective against meal antidepressant, and a subset of food category data ineffective against meal antidepressant; respectively calculating the mean value and the variance of the antidepressant indexes of foods in the two subsets; calculating the maximum threshold value a of the two subsets according to the mean value and the variance of the antidepressant indexes of the foods; constructing a Fisher linear discriminant function according to the maximum threshold value a; and (3) screening foods in the two sample subsets by adopting a Fisher linear discriminant function to obtain foods with depression diet indexes.
Preferably, the clustering process of the selected food categories includes: the obtained cluster analysis sample is the number of foods for screening out depression diet indexes; calculating the March distance d ij between every two sample points; calculating the sum of squares of the dispersion among all samples according to the mahalanobis distance; constructing a cluster map according to the square sum of the distance differences; and clustering all the samples according to the clustering graph to obtain foods of different categories.
Preferably, the process of setting meal recommendation results includes:
Step 1: setting food nodes a=v 11、v12、v13、...v1p =z, each node representing a food; weight w (v 1i,v1j)＝wij >0, weight w ij indicates that eating the i, j food can improve the patient's "value" of depression;
Step2: selecting i=k type food data;
Step 3: setting the food quantity of depressed patients;
step 4: calculating the shortest path from node a to node b by adopting a shortest path algorithm according to the physical health condition information of the user and the set food quantity of the depressed patient; wherein the path is a depression index value of the user;
Step 5: judging whether the current calculated path is the shortest path or not, and judging whether the value of the current food type i is equal to 3 or not; outputting all the shortest path sets which are the optimal meal recommendation results; otherwise, the food category i is added with 1, and the step 2 is returned.
The invention has the beneficial effects that:
According to the invention, a digital health management database based on depression meal indexes is constructed, and meal intelligent recommendation query is directly carried out in the database through data of a user, so that the accuracy of meal management of the user is enhanced; according to the Fisher discriminant analysis method, the depression diet indexes of the foods are analyzed and screened, so that the food diet recommendation result is more accurate; according to the invention, the influence of different foods on the depression meal index is calculated through the body data of the user and the shortest path algorithm, so that the foods are recommended in a targeted manner, and the accurate health management of the user is realized.
Drawings
FIG. 1 is a system flow diagram of the present invention;
FIG. 2 is a graph of the mean of the Sobel operator of the present invention;
FIG. 3 is a flow chart of the segmentation of food images according to the present invention;
FIG. 4 is a flow chart of the present invention for conducting a depression meal index analysis;
FIG. 5 is a flow chart of cluster analysis in accordance with the present invention;
FIG. 6 is a flow chart of the shortest path algorithm of the present invention;
FIG. 7 is a flow chart of evaluation analysis according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
A depression meal management method based on food image segmentation, as shown in fig. 1, the method comprising: constructing a digital health management database based on depression meal indexes; acquiring physical health condition data of a current user; according to the physical health condition data of the current user, carrying out intelligent data query and reasoning by combining a digital health management database based on the depression meal index to obtain a meal recommendation result of the current user;
the process of constructing a digital health management database based on depression meal indices includes:
s1: acquiring physical health condition data information and food pictures of a user;
S2: dividing the food pictures to obtain each food picture;
S3: performing depression diet index analysis on the separated food images by using a Fisher discriminant analysis method, and screening the number of foods with depression diet indexes;
S4: clustering the screened foods;
s5: setting a meal recommendation result according to the physical health condition information of the user and the clustered food category, and storing the physical health condition data information of the user and the meal recommendation result into a database.
As shown in fig. 3, segmenting the food image includes preprocessing the food image, the preprocessing including: filtering, enhancing and complementing the image; converting the preprocessed picture into a matrix; and performing edge segmentation on the matrix by adopting a Robert operator and a Sobel operator, and outputting an image and the matrix after the edge segmentation.
Specifically, an edge-based segmentation algorithm is adopted for image segmentation, and commonly used segmentation operators include Robert and Sobel operators. The Robert operator uses a2×2 template as a relatively simple first-order differential operator, has small calculation amount and sensitivity to detail reflection, and can give relatively thinner edges relative to other 3×3 operators, so that the maximum advantage is that the calculation speed is very high. The Sobel operator is a3×3 first order difference operator used to calculate the gray approximation of the image luminance function, and the use of this operator at any point in the image will produce the corresponding gray vector or its normal vector.
The Robert operator is a gradient calculation method of oblique deviation difference, the gradient size represents the strength of the edge, and the gradient direction is orthogonal with the edge trend. The Robert operator has accurate edge positioning and is extremely sensitive to noise. The gradient operator is defined as:
wherein G (x, y) represents an output image, Representing the x-direction gradient, f (x, y) representing the input image,/>Representing the y-direction gradient.
The operator can be simplified into:
from the reduction result, a diagonal Robert operator of image discretization (differential substitution partial derivative) can be obtained:
As shown in fig. 2, the roll factors of Sobel are divided into two groups: the 3 x 3 matrices of G x and G y represent the lateral and longitudinal directions, respectively, and the luminance difference approximation of the lateral and longitudinal directions can be obtained by convolving the matrices with the image plane. If the original image is denoted by a, G x and G y represent the image gray values for the lateral and longitudinal edge detection, respectively. The expression is as follows:
and/>
Gx＝(-1)·f(x-1,y-1)+0·f(x,y-1)+1·f(x+1,y-1)+(-2)·f(x-1,y)+0·f(x,y)+2·f(x+1,y)+(-1)·f(x-1,y+1)+0·f(x,y+1)+1·f(x+1,y+1)
Gy＝1·f(x-1,y-1)+2·f(x,y-1)+1·f(x+1,y-1)+0·f(x-1,y)+0·f(x,y)+0·f(x+1,y)+(-1)·f(x-1,y+1)+(-2)·f(x,y+1)+(-1)·f(x+1,y+1)
where f (a, b) represents the gray value of the point (a, b) of the image.
The lateral and longitudinal gray values of each pixel of the image, the gray size of the point is calculated by the following formula:
to improve efficiency, use the approximation of the non-square:
|G|＝|Gx|+|Gy|
if the gradient G is greater than a certain threshold, then the point is considered to be an (x, y) edge point.
After calculation by the selected operator, two formats of output are carried out on the sample: a segmented image and segmented pixel values.
As shown in fig. 4, the split food images were subjected to depression meal index analysis using Fisher discriminant analysis. The process comprises the following steps: all foods were divided into two sample subsets, two subsets being: a subset of food category data effective against meal antidepressant and a subset of food category data ineffective against meal antidepressant; respectively calculating the mean value and the variance of the antidepressant indexes of foods in the two subsets; calculating the maximum threshold value a of the two subsets according to the mean value and the variance of the antidepressant indexes of the foods; constructing a Fisher linear discriminant function according to the maximum threshold value a; and (3) screening foods in the two sample subsets by adopting a Fisher linear discriminant function to obtain foods with depression diet indexes.
The method specifically comprises the following steps: after the food is subjected to image segmentation, the food needs to be judged to see which type the food belongs to; fisher discrimination is adopted in the multivariate statistics, and discrimination analysis is carried out on the overall sample. Fisher's discrimination considers only two general cases, namely whether it is effective for depression index. Let p-dimensional populations X 1 and X 2 both have second order distances. In the case of the p-dimension, the linear combination of x, y=a T x. That is, the average value of y is:
μy1＝E(y|y＝aTx,x∈X1)＝aTμ1,
μy2＝E(y|y＝aTx,x∈X2)＝aTμ2,
Wherein the mean of X 1,X2 is μ 1,μ2, the covariance matrix is Σ (Σ > 0), μ y1 represents the mean of the first kind of linear combination, E represents the expectation, a T represents the p-dimensional real vector, X represents the variable, X 1 and X 2 represent the variables belonging to the first and second kinds, μ 1 represents the mean of X1, and μ y2 represents the mean of the second kind of linear combination.
The variance is:
wherein, The variance of the linear combination of variables x is expressed, and Var is the variance of the linear combination of x.
Calculating a ratio according to the variance and the mean value, wherein the expression is as follows:
Where δ=μ 1-μ2 is the mean vector difference of the two populations, only a has to be chosen such that the above ratio is maximized.
Maximum when a=cΣ -1 δ (c+.0 and constant) is chosen,
When c=1, the linear function y=a Tx＝(μ1-μ2)TΣ-1 x is a Fisher linear discriminant function, further letting:
Then mu y1-K>0,μy2 -K <0.
From the above expression, it can be seen that: fisher discriminant rules are:
And (3) making:
The discrimination rule is:
When the overall parameters are known, μ, Σ is estimated using the samples.
As shown in fig. 5, the process of clustering the selected food categories includes: the obtained cluster analysis sample is the number of foods for screening out depression diet indexes; calculating the March distance d ij between every two sample points; calculating the sum of squares of the dispersion among all samples according to the mahalanobis distance; constructing a cluster map according to the sum of squares of the distance differences; and clustering all the samples according to the clustering graph to obtain foods of different categories.
The module mainly uses clustering analysis to cluster foods, such as clustering foods into three categories of excellent, medium and no influence on depression. The clusters commonly used are: shortest and longest distance method clustering, K-means clustering, fuzzy clustering, dispersion square sum clustering, R-type clustering, Q-type clustering and the like; different clustering methods are available for different data. Wherein analyzing the food data includes food containing a plurality of nutritional indicators, and depressed patient characteristics represented by a plurality of variables.
The Q-type cluster analysis used is mainly to classify foods by a quantitative method and describe the degree of similarity between foods. Wherein the food has multidimensional index. For food samples to be classified, p indices are required for the descriptive description. Each sample point may be a point in the p-dimensional space. Thus, the "distance" can be used to measure the similarity of each sample point.
Let Ω be the sample point set, the distance d (x, y) be a function of Ω×Ω→r +, and the following condition is satisfied:
(1)d(x,y)≥0，x,y∈Ω；
(2) d (x, y) =0, if and only if x=y;
(3)d(x,y)＝d(y,x),x,y∈Ω；
(4)d(x,y)≤d(x,z)+d(z,y),x,y,z∈Ω.
the clustering analysis mainly uses the mahalanobis distance, and the formula is as follows:
Where x, y is from the observations of the p-dimensional ensemble Z; sigma is a covariance matrix, often requiring sample estimation; the mahalanobis distance is constant for all linear transformations and is not affected by the dimension. In addition, "distance" is also a coordinate distance, chebyshev distance, mahalanobis distance, or the like.
Sum of squares of dispersion:
Then there are: d (G 1,G2)＝D12-D1-D2, clustering ensures that G 1 and G 2 are very small in internal distance, but very large in total distance for both classes, where D 1、D2、D12 represents the sum of the squares of the groups, the total squares, x i represents the group 1 element, Represents group 1 mean, T represents transpose, G 1 represents sample population from the first group, G 2 represents sample population from the second group, x j represents group 2 element,/>Represents the second set of element means, x k represents the total element,/>The total mean value is represented by n 1, the number of samples G 1, and n 2, the number of samples G 2.
As shown in fig. 6, the process of setting meal recommendation results includes:
Step 1: setting food nodes a=v 11、v12、v13、...v1p =z, each node representing a food; weight w (v 1i,v1j)＝wij >0, weight w ij indicates that eating the i, j food can improve the patient's "value" of depression;
Step2: selecting i=k type food data;
Step 3: setting the food quantity of depressed patients;
step 4: calculating the shortest path from node a to node b by adopting a shortest path algorithm according to the physical health condition information of the user and the set food quantity of the depressed patient; wherein the path is a depression index value of the user;
Step 5: judging whether the current calculated path is the shortest path or not, and judging whether the value of the current food type i is equal to 3 or not; outputting all the shortest path sets which are the optimal meal recommendation results; otherwise, the food category i is added with 1, and the step 2 is returned.
Specifically, the path of the connected graph represents the increased value of depression index values from the current node to the reachable node (food). The Dijkstra algorithm is to find the shortest path (depression index value) from node a (food a) to node b (food b) in a weighted simple connected graph. Dividing all nodes in the graph into two groups, wherein each node corresponds to a distance value; the first group includes nodes of the shortest path that have been determined, the value corresponding to a node being the shortest path length (minimum depression index value) from v 0 to that node; the second group comprises nodes of which the shortest path is not yet determined, and the distance value corresponding to the nodes is the shortest distance from v 0 to the nodes of the first group through the nodes (intermediate nodes); the nodes of the second group are added to the first group in order of increasing shortest path length until all nodes reachable by v 0 are included in the first group. During this process, it is always maintained that: the shortest path length from v 0 to each node of the first set is no greater than the path length from v 0 to any node of the second set.
Setting a source point as v 0, and enabling v 0 to enter a first group initially, wherein the distance value of v 0 is 0; the second group contains all other nodes, and the distance values corresponding to the nodes are defined as follows:
Each time a node v m with the smallest distance is selected from the nodes of the second group and added to the first group, each time a node v m is added to the first group, the distance between the nodes of the second group is corrected.
If the added node v m makes the distance v 0 to v i shorter (d 0i>d0m+wmi), the distance v i is modified (d 0i←d0m+wmi). And after correction, selecting a node with the smallest distance to add into the first group until the first group contains all the needed nodes or no node can be added again.
An embodiment of diet recommendation based on weighted graph, the method comprising: let G have nodes a=v 11,v12,v13,...v1p =z, each node representing a food; the weighting w (v 1i,v1j)＝wij >0, weight w ij indicates that eating the i, j foods can make the patient improve the depression "value". If e (v 1i,v1j) is not the side of G, i.e., eating the i, j foods is unlikely to occur, then w ij = infinity:
step one: a first class v 11,v12,...,v1p analyzed by clusters;
Step two: let l (u 0)＝0,l(vi) = infinity, i=1, 2,..n,
Step three: if it isLetting u take the node which does not belong to S and has the smallest l (u), letting S=S { u }, and repeatedly executing the second step;
Step four: for all nodes v not belonging to S, let l (v) =l (u) +w (u, v) if l (u) +w (u, v) < l (v);
step five: i (z) is the shortest length of nodes a through z, and is also the optimal diet.
Step six: listing all nodes with the shortest as set β= { β 1,β2,...,βk };
Step seven: if the actual situation is insufficient in nutrition, returning to the first step, selecting the foods of the second type of clusters as nodes, and executing the second to sixth steps.
In the embodiment, the module realizes that a rank sum ratio comprehensive evaluation method is used, the food recommendation module and the doctor suggestion are subjected to comparative analysis, and a dimensionless statistic Rank Sum Ratio (RSR) of the food recommendation module and the doctor suggestion is calculated for sequencing; the comprehensive evaluation method of the rank sum ratio is that in a matrix of n rows and m columns, dimensionless statistic RSR is obtained through rank conversion; and (5) directly sequencing or grading the merits of the evaluation objects by using the RSR value.
The method specifically comprises the following steps: let x (1),x2,...,x(n) be the order of samples with sample size n from small to large. If x i＝x(k), k is the rank in the sample, denoted R i, and for each i=1. R 1,R2,...,Rn is collectively referred to as rank statistic. As shown in fig. 7, the steps are as follows:
Step one: combining the scheme beta and the doctor scheme into a matrix A of n rows and m columns;
Step two: rank-ordering; the ranks of each food scheme in the matrix A are compiled, wherein the benefit type index (food) is compiled from small to large, the cost type index (food) is compiled from large to small, and the average ranks are compiled by the same index (food) data. The resulting matrix is denoted r= (R ij)n×m.
Step three: calculating a Rank Sum Ratio (RSR), and calculating a Weighted Rank Sum Ratio (WRSR) when the weights of the various indexes are different; where w j is the weight of the j-th evaluation index (food),
Wherein w j represents the weight of the jth index, RSR i represents the rank sum ratio of the jth evaluation index, m represents the number of indexes, n represents the column element of the current index, R ij represents the R matrix element, WRSR i represents the weighted rank sum ratio.
Step four: sequencing in steps; the evaluation objects are ranked in steps according to the value of RSR i or WRSR i (i=1, 2,., n), and the larger the value is, the better the effect is.
While the foregoing is directed to embodiments, aspects and advantages of the present invention, other and further details of the invention may be had by the foregoing description, it will be understood that the foregoing embodiments are merely exemplary of the invention, and that any changes, substitutions, alterations, etc. which may be made herein without departing from the spirit and principles of the invention.
Claims (6)
1. A depression meal management method based on food image segmentation, comprising: constructing a digital health management database based on depression meal indexes; acquiring physical health condition data of a current user; according to the physical health condition data of the current user, carrying out data query analysis on a digital health management database based on the depression meal index to obtain a meal recommendation result of the current user;
the process of constructing a digital health management database based on depression meal indices includes:
s1: acquiring physical health condition data information and food pictures of a user;
S2: dividing the food pictures to obtain each food picture;
S3: constructing a depression meal index by using a Fisher discriminant analysis method, analyzing the depression meal index of the segmented food image, and screening out food pictures for controlling the depression meal index; the analysis of depression meal index of the segmented food image by Fisher discriminant analysis comprises: all foods were divided into two sample subsets, two subsets being: a subset of food category data effective against meal antidepressant and a subset of food category data ineffective against meal antidepressant; respectively calculating the mean value and variance of the antidepressant indexes of foods in the two data subsets; calculating the maximum threshold value a1 of the two subsets according to the mean value and the variance of the antidepressant indexes of the foods; constructing a Fisher linear discriminant function according to the maximum threshold value a1; screening foods in the two sample subsets by adopting Fisher linear discriminant function to obtain food pictures for controlling depression diet indexes;
s4: clustering the screened food pictures;
S5: setting a meal recommendation result according to the physical health condition information of the user and the clustered food category, and storing the physical health condition data information of the user and the meal recommendation result into a database;
The process of setting meal recommendation results comprises the following steps:
Step 1: setting food nodes a, wherein each node represents a food; a weight w (v 1i,v1j)＝wij >0, weight w ij representing a depression index value that improves patient outcome by eating the i, j foods;
Step2: selecting i=k type food data;
Step 3: setting the food quantity of depressed patients;
Step 4: calculating a shortest path from the node a to the node b by adopting a shortest path algorithm according to the physical health condition information of the user and the set food quantity of the depressed patient;
Step 5: judging whether the current calculated path is the shortest path or not, and judging whether the value of the current food type i is equal to 3 or not; determining whether the shortest path is a minimum value for determining a depression index value of the user; outputting all the shortest path sets which are the optimal meal recommendation results; otherwise, the food category i is added with 1, and the step 2 is returned.
2. The depression meal management method based on food image segmentation according to claim 1, wherein the process of segmentation processing of food pictures comprises: preprocessing the food image, wherein the preprocessing comprises filtering, enhancing and complementing the image; converting the preprocessed picture into a matrix; and performing edge segmentation on the matrix by adopting a Robert operator and a Sobel operator, and outputting an image and the matrix after the edge segmentation.
3. The depression meal management method based on food image segmentation according to claim 1, wherein the expression of Fisher's linear discriminant function is:
Wherein u represents a sample to be discriminated of food, mu 1 represents an anti-depression index mean value of a subset of food category data effective for meal anti-depression, mu 2 represents an anti-depression index mean value of a subset of food category data ineffective for meal anti-depression, and Σ represents a covariance matrix.
4. The method of claim 1, wherein clustering the selected food pictures comprises: obtaining a cluster analysis sample, wherein the sample is a food picture for screening out depression diet indexes; calculating the March distance d ij between every two sample points; calculating the sum of squares of the distance differences between all samples according to the mahalanobis distance; constructing a cluster map according to the square sum of the distance differences; and clustering all the samples according to the clustering graph to obtain foods of different categories.
5. The method for managing a depressive disorder diet based on food image segmentation according to claim 4, wherein the equation for calculating the mahalanobis distance between each sample point is:
Where m and n each represent p-dimensional observations in the population of the diet, Σ represents the covariance matrix, and T represents the transpose.
6. The method for meal management of depression based on segmentation of food images according to claim 4, wherein the formula for calculating the sum of squares of the distance differences between all samples is:
Wherein D 1、D2、D12 represents the sum of the inter-group squares, the intra-group squares, and the total squares, x i represents the group 1 element, Represents group 1 mean, T represents transpose, G 1 represents sample population from the first group, G 2 represents sample population from the second group, x j represents group 2 element,/>Represents the second set of element means, x k represents the total element,/>The total mean value is represented by n 1, the number of samples G 1, and n 2, the number of samples G 2.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211628844.1A CN115798685B (en) | 2022-12-18 | 2022-12-18 | Depression diet management method based on food image segmentation |
| PCT/CN2023/136614 WO2024131524A1 (en) | 2022-12-18 | 2023-12-06 | Depression diet management method based on food image segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211628844.1A CN115798685B (en) | 2022-12-18 | 2022-12-18 | Depression diet management method based on food image segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115798685A CN115798685A (en) | 2023-03-14 |
| CN115798685B true CN115798685B (en) | 2024-05-24 |
Family
ID=85425578
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211628844.1A Active CN115798685B (en) | 2022-12-18 | 2022-12-18 | Depression diet management method based on food image segmentation |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115798685B (en) |
| WO (1) | WO2024131524A1 (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116453653B (en) * | 2023-06-16 | 2023-09-01 | 北京四海汇智科技有限公司 | Meal data set construction and query method and device based on automatic sampling technology |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011080957A (en) * | 2009-10-09 | 2011-04-21 | Ajinomoto Co Inc | Fibromyalgia and depression evaluation method, fibromyalgia and depression evaluation apparatus, and fibromyalgia and depression evaluation method |
| CN104765980A (en) * | 2015-05-04 | 2015-07-08 | 哈尔滨理工大学 | Intelligent diet assessment method based on cloud computing |
| CN104778374A (en) * | 2015-05-04 | 2015-07-15 | 哈尔滨理工大学 | Automatic dietary estimation device based on image processing and recognizing method |
| KR20170030334A (en) * | 2015-09-09 | 2017-03-17 | 연세대학교 산학협력단 | Apparatus and Method of Enhancing Visual Flavor of Food Image |
| WO2020132478A1 (en) * | 2018-12-20 | 2020-06-25 | Dqpn, Llc | Diet quality fingerprinting |
| CN115130006A (en) * | 2022-08-04 | 2022-09-30 | 北京富通智康科技有限公司 | User portrayal method based on health management label |
| CN115171848A (en) * | 2022-07-25 | 2022-10-11 | 重庆邮电大学 | Intelligent diet recommendation system based on food image segmentation and uric acid index |
| CN115315752A (en) * | 2020-04-27 | 2022-11-08 | 雀巢产品有限公司 | System and method for providing individualized recommendations for a healthy microbiome |
| CN115424702A (en) * | 2022-09-06 | 2022-12-02 | 南华大学 | Food matching intelligent management system |
| CN115428094A (en) * | 2020-04-17 | 2022-12-02 | 派弗股份有限公司 | Improved method for determining glycemic response |
-
2022
- 2022-12-18 CN CN202211628844.1A patent/CN115798685B/en active Active
-
2023
- 2023-12-06 WO PCT/CN2023/136614 patent/WO2024131524A1/en unknown
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011080957A (en) * | 2009-10-09 | 2011-04-21 | Ajinomoto Co Inc | Fibromyalgia and depression evaluation method, fibromyalgia and depression evaluation apparatus, and fibromyalgia and depression evaluation method |
| CN104765980A (en) * | 2015-05-04 | 2015-07-08 | 哈尔滨理工大学 | Intelligent diet assessment method based on cloud computing |
| CN104778374A (en) * | 2015-05-04 | 2015-07-15 | 哈尔滨理工大学 | Automatic dietary estimation device based on image processing and recognizing method |
| KR20170030334A (en) * | 2015-09-09 | 2017-03-17 | 연세대학교 산학협력단 | Apparatus and Method of Enhancing Visual Flavor of Food Image |
| WO2020132478A1 (en) * | 2018-12-20 | 2020-06-25 | Dqpn, Llc | Diet quality fingerprinting |
| CN115428094A (en) * | 2020-04-17 | 2022-12-02 | 派弗股份有限公司 | Improved method for determining glycemic response |
| CN115315752A (en) * | 2020-04-27 | 2022-11-08 | 雀巢产品有限公司 | System and method for providing individualized recommendations for a healthy microbiome |
| CN115171848A (en) * | 2022-07-25 | 2022-10-11 | 重庆邮电大学 | Intelligent diet recommendation system based on food image segmentation and uric acid index |
| CN115130006A (en) * | 2022-08-04 | 2022-09-30 | 北京富通智康科技有限公司 | User portrayal method based on health management label |
| CN115424702A (en) * | 2022-09-06 | 2022-12-02 | 南华大学 | Food matching intelligent management system |
Non-Patent Citations (1)
| Title |
|---|
| 基于改进人工蜂群模糊聚类的葡萄图像快速分割方法;罗陆锋 等;农业机械学报;第46卷(第03期);第23-28页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115798685A (en) | 2023-03-14 |
| WO2024131524A1 (en) | 2024-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hancer et al. | A survey on feature selection approaches for clustering | |
| Erisoglu et al. | A new algorithm for initial cluster centers in k-means algorithm | |
| CN110929029A (en) | Text classification method and system based on graph convolution neural network | |
| Dzulkalnine et al. | Missing data imputation with fuzzy feature selection for diabetes dataset | |
| Drab et al. | Clustering in analytical chemistry | |
| Long et al. | Boosting and microarray data | |
| Abdalameer et al. | A new validity clustering index-based on finding new centroid positions using the mean of clustered data to determine the optimum number of clusters | |
| CN113344075A (en) | High-dimensional unbalanced data classification method based on feature learning and ensemble learning | |
| CN115798685B (en) | Depression diet management method based on food image segmentation | |
| Chen et al. | A novel information-theoretic approach for variable clustering and predictive modeling using dirichlet process mixtures | |
| Villmann et al. | Classification of mass-spectrometric data in clinical proteomics using learning vector quantization methods | |
| Purushotham et al. | A comparative study of RIFCM with other related algorithms from their suitability in analysis of satellite images using other supporting techniques | |
| Miglio et al. | The comparison between classification trees through proximity measures | |
| CN109885712B (en) | Pulmonary nodule image retrieval method and system based on content | |
| Kılıçarslan et al. | A comparative study of bread wheat varieties identification on feature extraction, feature selection and machine learning algorithms | |
| Sebag et al. | ROC-based evolutionary learning: Application to medical data mining | |
| Rodosthenous et al. | Multi-view data visualisation via manifold learning | |
| CN113688906A (en) | Customer segmentation method and system based on quantum K-means algorithm | |
| CN117195027A (en) | Cluster weighted clustering integration method based on member selection | |
| Gupta et al. | Feature importance for human epithelial (HEp-2) cell image classification | |
| Krızek | Feature selection: stability, algorithms, and evaluation | |
| Abouabdallah et al. | Does clustering of DNA barcodes agree with botanical classification directly at high taxonomic levels? Trees in French Guiana as a case study | |
| Shaik et al. | Fuzzy-adaptive-subspace-iteration-based two-way clustering of microarray data | |
| Elsten et al. | SUBiNN: a stacked uni-and bivariate k NN sparse ensemble | |
| Braiki et al. | Fuzzy model for the automatic recognition of human dendritic cells |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240326 Address after: Room 801, 85 Kefeng Road, Huangpu District, Guangzhou City, Guangdong Province Applicant after: Guangzhou Dayu Chuangfu Technology Co.,Ltd. Country or region after: China Address before: 400065 Chongwen Road, Nanshan Street, Nanan District, Chongqing Applicant before: CHONGQING University OF POSTS AND TELECOMMUNICATIONS Country or region before: China |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240428 Address after: 530000 ASEAN International Cell Bank, No. 32 Yuxiang Road, Nanning Area, China (Guangxi) Pilot Free Trade Zone, Nanning City, Guangxi Zhuang Autonomous Region Applicant after: Guangxi kangjiu Biotechnology Co.,Ltd. Country or region after: China Address before: Room 801, 85 Kefeng Road, Huangpu District, Guangzhou City, Guangdong Province Applicant before: Guangzhou Dayu Chuangfu Technology Co.,Ltd. Country or region before: China |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |