Computer-implemented medical analytics method and system employing a modified mini-max procedure
First Claim
1. A medical analytics method implemented on a computer for diagnosing at least one disease (i) afflicting a patient based on clinical data (m) that excludes subjective qualities of said clinical data (m) and prevalence of said at least one disease (i), said method comprising:
- a) compiling a knowledge base of disease (i) models exhibiting said clinical data (m);
b) inputting clinical data present (j) in said patient into said computer;
c) matching said clinical data present (j) with said clinical data (m) in said knowledge base;
d) composing a differential diagnosis list of ruled in diagnoses (k) for each of said disease (i) models exhibiting at least one clinical datum (m) matching at least one clinical datum present (j);
e) computing a probability P(k) for each of said ruled in diagnoses (k) by a mini-max procedure comprising;
1) obtaining sensitivities S(i)m of each of said clinical data (m) for diseases (i) based on disease (i) models that comprise a total number of disease (i) cases;
2) computing positive predictive values PP(k)j for clinical data present (j) supporting each of said ruled in diagnoses (k) as follows;
where S(k)j are sensitivities of each clinical datum present (j) to said diagnoses (k), and n is the number of said ruled in diagnoses (k);
3) assigning as probability P(k) of each ruled in diagnosis (k) the maximum value among said positive predictive values PP(k)1 through PP(k)z;
P(k)=max(PP(k)1, PP(k)2, . . . , PP(k)2),where z is the number of said ruled in diagnoses (j); and
f) displaying said differential diagnosis list and said probability P(k) for each diagnosis (k) on a display for medical analytics.
1 Assignment
0 Petitions
Accused Products
Abstract
A method and system for medical analytics implemented on a computer and designed to aid a medical professional in diagnosing one or more diseases afflicting a patient. In contrast to prior art, the present method is based on using clinical data (m) that excludes subjective qualities of and also excludes prevalence of the one or more diseases (i). The method uses a knowledge base that contains disease (i) models exhibiting clinical data (m). Clinical data present (j) in the patient are input into the computer. Then, clinical data present (j) are matched with clinical data (m) in the knowledge base to enable the computer to compose a differential diagnosis list of ruled in diagnoses (k), where k=1 . . . n, for each of the disease (i) models that exhibits at least one clinical datum (m) that matches at least one clinical datum present (j) in the patient. In a key step, the computer computes a probability P(k) for each of the ruled in diagnoses (k) with the aid of a mini-max procedure that overcomes prior art limitations of the Bayes formulation and permits the analytics method to consider concurrent and competing diagnoses (k). Furthermore, the method composes pairs of clinical data present (j) and absent (r) in the patient to aid the medical professional in evaluating diagnoses and determining the most cost-effective clinical data to collect for conducting an effective and rapid diagnostic quest.
24 Citations
20 Claims
-
1. A medical analytics method implemented on a computer for diagnosing at least one disease (i) afflicting a patient based on clinical data (m) that excludes subjective qualities of said clinical data (m) and prevalence of said at least one disease (i), said method comprising:
-
a) compiling a knowledge base of disease (i) models exhibiting said clinical data (m); b) inputting clinical data present (j) in said patient into said computer; c) matching said clinical data present (j) with said clinical data (m) in said knowledge base; d) composing a differential diagnosis list of ruled in diagnoses (k) for each of said disease (i) models exhibiting at least one clinical datum (m) matching at least one clinical datum present (j); e) computing a probability P(k) for each of said ruled in diagnoses (k) by a mini-max procedure comprising; 1) obtaining sensitivities S(i)m of each of said clinical data (m) for diseases (i) based on disease (i) models that comprise a total number of disease (i) cases; 2) computing positive predictive values PP(k)j for clinical data present (j) supporting each of said ruled in diagnoses (k) as follows; where S(k)j are sensitivities of each clinical datum present (j) to said diagnoses (k), and n is the number of said ruled in diagnoses (k); 3) assigning as probability P(k) of each ruled in diagnosis (k) the maximum value among said positive predictive values PP(k)1 through PP(k)z;
P(k)=max(PP(k)1, PP(k)2, . . . , PP(k)2),where z is the number of said ruled in diagnoses (j); and f) displaying said differential diagnosis list and said probability P(k) for each diagnosis (k) on a display for medical analytics. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
3. The method of claim 2, further comprising:
-
k) repeating the steps g) through j) to create additional clinical data pairs (j, r); and l) computing partial probabilities P(k)j,r for said additional clinical data pairs (j, r) such that said partial probabilities P(1)j,r, . . . P(k)j,r, P(n)j,r for corresponding ruled in diagnoses (k), where k=1 . . . n, satisfy a normalization condition;
-
-
4. The method of claim 3, wherein said mini-max procedure further comprises creating a mini-max table for each of said diagnoses (k) in said differential diagnosis list, whereby a first data column of each said mini-max table comprises said positive predictive values PP(k)j of said clinical data present (j) in said patient for said diagnosis (k) for which said mini-max table is created and a first row of each said mini-max table comprises said sensitivities S(k)r of said clinical data absent (r) for said diagnosis (k) for which said mini-max table is created.
-
5. The method of claim 4, further comprising the steps of:
-
a) transferring each said partial probability P(k)j,r into cells of said mini-max table where said PP(k)j for each said clinical datum present (j) and said sensitivity S(k)r for each said clinical datum absent (r) converge; and b) selecting from among said partial probabilities P(k)j,r in said cells a determining partial probability DP(k)j,r having the smallest value in its row and the greatest value in its column.
-
-
6. The method of claim 5, wherein from each said mini-max table, said determining partial probability DP(k)j,r is selected as a total probability TP(k) for said diagnosis (k) for which said mini-max table is created.
-
7. The method of claim 6, further comprising confirming as final diagnoses (k) those of said diagnoses (k) for which said total probability TP(k) is greater than a confirmation threshold value CT, and ruling out those of said diagnoses (k) for which said total probability TP(k) is smaller than a deletion threshold DT.
-
8. The method of claim 7, further comprising ending said method when all said diagnoses (k) in said differential diagnosis list have satisfied said confirmation threshold CT and said deletion threshold DT.
-
9. The method of claim 1, further comprising recommending a best cost-benefit clinical datum (m) to investigate in said patient by:
-
a) selecting from said disease (i) models corresponding to said diagnoses (k) ruled in into said differential diagnosis list a clinical datum not yet investigated (y) in said patient; and b) determining a cost C(y) of collecting said clinical datum not yet investigated (y) as follows;
C(y)=max(expense(y),risk(y),discomfort(y)),where said cost C(y) comprises the maximum of expense, risk and discomfort for said patient.
-
-
10. The method of claim 9, further comprising computing a total probability TP(k) for each of said diagnoses (k) ruled in into said differential diagnosis list by including in said mini-max procedure said clinical datum not yet investigated (y).
-
11. The method of claim 10, wherein said clinical datum not yet investigated (y) is treated as if present in said patient when computing said total probability TP(k) and wherein said clinical datum not yet investigated (y) is treated as if absent in said patient when computing said total probability TP(k).
-
12. The method of claim 11, further comprising selecting said clinical datum not yet investigated (y) such that said cost C(y) is minimized and a change to said total probability TP(k) is maximized.
-
13. The method of claim 10, further comprising simultaneously recommending collecting from said patient a plurality of clinical data not yet investigated (y) that minimize said cost C(y) and maximize a change to said total probability TP(k).
-
14. The method of claim 13, further comprising computing a total probability TP(k) for each of said diagnoses (k) in said differential diagnosis list with said plurality of clinical data not yet investigated (y) treated as if present in said patient, and then computing said total probability TP(k) with said plurality of clinical data not yet investigated (y) treated as if absent in said patient.
-
15. The method of claim 1, further comprising determining whether the absence of any clinical datum among said clinical data (m) will decrease a total probability TP(k) of said corresponding diagnosis (k), and aborting said mini-max procedure for said clinical datum (m) when its absence does not decrease said total probability TP(k).
-
16. The method of claim 1, further comprising the steps of:
-
a) distinguishing competitive diagnoses (k) from concurrent diagnoses (k); and b) applying said distinction to diagnoses (k) ruled in on said differential diagnosis list.
-
-
17. The method of claim 1, further comprising diagnosing complex clinical presentations comprising the steps of:
-
a) creating complex clinical presentation models listing all potentially related diagnoses (k); and b) transferring all said potentially related diagnoses (k) to said differential diagnosis list when one confirmed diagnosis matches any of said potentially related diagnoses.
-
-
18. The method of claim 1, further comprising checking for interactions between drugs having the potential of masking any of said clinical data (m) belonging to a primary disease (i), comprising the steps of:
-
a) flagging all said disease (i) models in said knowledge base that include at least one clinical datum (m) susceptible to being masked by said drugs; b) listing for said disease (i) models all said drugs and concurrent diagnoses (k) having the potential of masking any of said clinical data (m) belonging to said primary disease (i); c) confirming presence of any of said drugs in said patient; and d) removing any clinical datum (m) masked by said drugs from consideration if at least one of said drugs is confirmed present.
-
-
19. The method of claim 1, further comprising checking for interactions between concurrent diseases (k) having the potential of masking any of said clinical data (m) belonging to a primary disease (i), comprising the steps of:
-
a) flagging all said disease (i) models in said knowledge base that include at least one clinical datum (m) susceptible to being masked by said concurrent diseases (i); b) listing for said disease (i) models all said concurrent diseases (i) having the potential of masking any of said clinical data (m) belonging to said primary disease (i); c) confirming presence of any of said concurrent diseases (i) in said patient; and d) removing any clinical datum (m) masked from each corresponding disease (i) that contains said clinical datum (m) in its disease (i) model.
-
-
20. A medical analytics system comprising a computer for aiding a physician in diagnosing at least one disease (i) afflicting a patient based on clinical data (m) that excludes subjective qualities of said clinical data (m) and prevalence of said at least one disease (i), said system comprising:
-
a) a knowledge base of disease (i) models exhibiting said clinical data (m); b) an input device for inputting clinical data present (j) in said patient into said computer for matching said clinical data present (j) with said clinical data (m) in said knowledge base; c) a display for displaying a differential diagnosis list of ruled in diagnoses (k) for each of said disease (i) models exhibiting at least one clinical datum (m) matching at least one clinical datum present (j); d) a processor for computing a probability P(k) for each of said ruled in diagnoses (k) by a mini-max procedure; e) a network connecting said processor with said knowledge base to enable said processor to obtain from said knowledge base sensitivities S(i)m of each of said clinical data (m) for diseases (i) based on disease (i) models that comprise a total number of disease (i) cases; and to further enable said processor to compute positive predictive values PP(k)j for clinical data present (j) supporting each of said ruled in diagnoses (k) as follows; where S(k)j are sensitivities of each clinical datum present (j) to said diagnoses (k), and n is the number of said ruled in diagnoses (k); and
to still further enable said processor to assign as probability P(k) of each ruled in diagnosis (k) the maximum value among said positive predictive values PP(k)1 through PP(k)z;
P(k)=max(PP(k)1, PP(k)2, . . . , PP(k)2),where z is the number of said ruled in diagnoses (j); whereby said display displays said differential diagnosis list and said probability P(k) for each diagnosis (k) to aid said physician in a diagnostic quest.
-
Specification