WO2002057990A2

WO2002057990A2 - Method for processing biological analysis data and expert biological analysis system therefor

Info

Publication number: WO2002057990A2
Application number: PCT/FR2002/000218
Authority: WO
Inventors: Patrick Rambaud; Lionel Chapy
Original assignee: Ippm Holding Sa
Priority date: 2001-01-19
Filing date: 2002-01-18
Publication date: 2002-07-25
Also published as: FR2819915A1; FR2819915B1; EP1412905A2; WO2002057990A3

Abstract

The invention concerns a method for processing biological analysis data comprising processing a set of data concerning the biological profile of a human or animal subject, in the form of biological items whereto are applied a set of rules providing conclusions in the form of statements in natural language. The method further comprises at least an operation which consists in assembling a set of conclusions resulting from a set of rules, so as to restrict the number of conclusions which are input in the next set of rules. Said method additionally comprises processing of genetic data concerning said patient in the form of a set of genetic items associated with a set of studied genes in said patient, said processing including executing genetic interpretation rules applied both to biological items and genetic items.

Description

"Process for processing biological analysis data and expert biological analysis system implementing this process"

The present invention relates to a method for processing biological analysis data. It also targets an expert biological analysis system.

In the field of biological analysis, there are already methods for determining the biological profile of a person from a set of measurements of characteristic physiological parameters. From these biological profiles and data relating to the person concerned such as his age, sex, physical condition, the practitioner can then make a set of conclusions leading to a diagnosis. A biological profile is in practice made up of a set of specific profiles, such as a protein profile or a lymphocyte typing.

The increase in the number of parameters involved in determining a biological profile makes it more and more difficult to draw coherent conclusions. To meet the expectations of practitioners prescribing biological analysis, expert biological analysis systems have been developed. These expert systems provide their user with processing of a set of items corresponding to biological measurement data and to personal data, with the provision of conclusions which can be directly exploited by the prescribing practitioner. The biological data processing methods implemented in these expert systems use a set of rules each applied to a determined combination of items from a global set of items corresponding to a set of measurements, examinations, assays performed on a patient or personal data. These rules lead to a set of conclusions which are previously drawn up by one or more expert practitioners. It turns out in practice that the number of possible theoretical conclusions in an expert biological analysis system which would aim to integrate a biological profile as complete as possible in the current state of the technologies available in biological analysis is so high that the feasibility of such an expert system and its implementation on conventional IT equipment other than large computing and storage machines could be called into question.

The aim of the present invention is to propose a method for processing biological analysis data which makes it possible, on the one hand, to effectively resolve the question of the volume of data to be processed and consequently makes such an expert system feasible, and which, on the other hand, provides the user practitioner with better relevance of the conclusions of interpretation of the analysis results.

This objective is achieved with a method for processing biological analysis data comprising processing a set of data relating to the biological profile of a human or animal subject, in the form of biological items to which a set of rules are applied. providing conclusions in the form of statements in natural language. This method is characterized in that the set of rules includes a plurality of groups of rules each associated with a group of analyzes from among a plurality of groups of analyzes constituting the biological profile and in that it further comprises at least an operation of grouping together a set of conclusions from a group of rules, so as to restrict the number of conclusions which are entered in a following group of rules.

The effect of such grouping operations is to make possible the implementation of an expert biological analysis system on personal office or portable computer equipment, without however affecting the precision and rigor of the conclusions supplied to the user. It should be noted that, in the present invention, it is indeed operations of grouping of conclusions and not of merger of chained rules as taught by document US 5,442,792 which discloses a compilation process for expert system.

Another object of the data processing method according to the invention is to allow the realization of an expert biological analysis system which integrates data of genetic profile whose knowledge is now essential for the diagnosis and treatment of a increasing number of ailments and pathologies.

This other object is achieved with a data processing method according to the invention characterized in that it further comprises processing of genetic data relating to this patient in the form of a set of genetic items associated with a set of genes studied in this patient, this treatment comprising the execution of rules of genetic interpretation applied to both biological items and genetic items. With this process combining interpretation of a biological profile and interpretation of a genetic profile, it becomes possible to propose more precise and more relevant conclusions due to the consideration of disorders linked to genes.

It should be noted that document WO01 / 1 6860 discloses an artificial intelligence system for a genetic analysis, which does not implement an operation of regrouping of a set of conclusions from a group of rules, as provided for the method according to the invention.

According to another aspect of the invention, an expert biological analysis system is proposed which implements the data processing method according to the invention.

Other advantages and characteristics of the invention will appear on examining the detailed description of a mode of implementation in no way limiting, and the appended drawings in which: - Figure 1 is a block diagram of an expert system biological analysis according to the invention, FIG. 2A illustrates an example of the internal structure of the engine of an expert system according to the invention, more particularly concerning rules of the inflammatory reaction, and

FIG. 2B illustrates an example of the internal structure of the engine of an expert system according to the invention, more particularly concerning rules of interpretation of the immunoglobulin. An expert system according to the invention of biological analysis according to the invention can in practice be implemented within a computer such as a desktop computer or a portable computer, and accessed locally or remotely. Its internal architecture, which can comply with current standards in terms of expert system, includes, with reference to FIG. 1, a module for collecting data for determining the respective biological (in particular protein) and genetic profiles of a patient, a module for collecting personalized information on the patient, rules of interpretation applied to a treatment of the biological profile carried out with a genetic interpretation, and an edition of conclusions usable by a user practitioner. The set of rules contained in this expert system according to the invention is organized into groups of rules each corresponding to a group of specific analyzes from among several groups of analyzes. For example, consider the rule group for inflammatory reaction or the rule group for interpreting immunoglobulins.

We will now describe an example of implementation of an expert system according to the invention, with reference to FIGS. 2A and 2B, limiting, for reasons of breadth of the description and clarity, to the only protein profile d 'A patient, it being understood that other specific biological profiles could be treated in an equivalent manner in the context of the present invention. A protein profile includes, in this expert system, optional items such as Item 43 = ROP or Item 45 = C4 and a set of mandatory items such as the following items:

Item 2 = Age Item 39 = TRF Item 3 ≈ Sex 39.1 = Normal Item 35 = ORO 39.2 = Increased (mandatory>

35.1 = Normal 119, not below)

35.2 = Increased 39.3 = Decreased

35.3 = Very increased Item 40 = ALB

35.4 = Decreased 40.1 = Normal or increased (> 89%)

Item 36 = HAPTO 40.2 = Decreased (<89%)

36.1 = Normal Item 41 = TRF / ALB

36.2 = Increased 41.1 = Normal

36.3 = Very increased 41.2 = Increased

36.4 = Decreased Item 42 = PAB

36.5 = Very decreased 42.1 = Normal or increased (> 84%)

36.6 = Hapto <10% 42.2 = Decreased (<84%)

Item 37 = CRP Item 44 = Electrophoresis of proteins

37.1 = Normal <33% completed

37.2 = Normal increased NO

37.3 = Increased IgM

37.4 = Very increased IgG

37.5 = Very very increased ... igA

Monoclonal peak neither M, nor G, nor A

Monoclonal double peak

Absence of monoclonal protein

The expert system according to the invention includes rules for the inflammatory reaction such as the following rules:

RINF1 = 35.1 +36.1 +37.1 = CINF1 RINF2 = 35.1 +36.1 + 37.2 = CINF2

RINF100 = 35.4 + 36.5 + 37.5 = CINF100

The conclusions associated with these rules of the inflammatory reaction are for example written as follows: CINF1 = No inflammatory reaction because the proteins of inflammation (CRP, Alpha-1-glycoprotein or orosomucoid, haptoglobin) are normal. CINF2 = The proteins of the inflammatory reaction are all within normal values. However, the level of CRP, although normal, may suggest the presence of microinflammations (to be taken into account in assessing the cardiovascular risk after formal elimination of any other potentially phlogogenic focus).

CINF100 = The results are in favor of an inflammatory process based solely on the sharp increase in CRP. This can be the start of an inflammatory phenomenon since CRP, the protein of acute inflammation, increases faster than other proteins of inflammation. Such induction is in favor of an infectious and / or inflammatory focus that is currently recent and very active, maintained in activity or in the reinduction phase. The low level of haptoglobin is in favor of hemolysis. Any haptoglobin result below 50% should be considered pathological. It may therefore be interesting to investigate the cause of this hemolysis. Here, the lowered level of I "alpha-1-glycoprotein should not evoke in the first place a drug treatment, but rather a protein leakage, or a hepatocyte insufficiency.

A table of normal values allows to associate with numerical values of items ORO, HPT, CRP, results such as:

We will now set out new rules for the inflammatory reaction corresponding to an interpretation of item 38 in relation to the conclusions of the three items 35, 36, 37.

RINF101 = CINF1 + 138.1 = CINF101 RINF102 = CINF1 + 138.2 = ^ CINF102 RINF103 = CINF1 + 138.3 => CINF103

RINF1 16 = CINF7 + 138.1 => (CINF1 16) = CINF101

RINF165 = CINF31 + 138.2 => IMPOSSIBLE

The set of conclusions corresponding to these rules includes for example:

CINF101 = No dissociation of the orosomucoid / haptoglobin couple which remains homogeneous. CINF102 = Dissociation between alpha-1-glycoprotein and haptogfobin. Dissociation is in favor of hemolysis. However, given the level of haptoglobin, hemolysis is not necessarily to be considered as pathological.

CINF107 = CINF101

CINF1400 = This decrease, with orosomucoid / haptoglobin dissociation, can result from a desialytating activity affecting the orosomucoid such as taking dibasic drugs modifying the antigenic structure of the protein (antibiotics, NSAIDs, beta-blockers, ...), or come from a slight loss of protein by leakage of urinary, digestive or skin origin because the orosomucoid, which is a protein of low molecular weight, is very sensitive to pathologies of leakage.

The start of the protein profile is thus presented as follows: - printing of the conclusion CINFI to CINF100 - printing of the conclusion CINF101 to CINF240

We then connect the conclusions CINF1 to CINF100 to items 39, 40, 41, 42. However, although there are in fact only 60 different conclusions, this would lead to far too many rules. It is therefore proposed to modify these 60 conclusions in order to arrive at a more limited number of conclusions, for example 6, with reference to FIG. 2A. If we consider 6 conclusions CINF301 to CINF306 related to items 39 (TRF), 40 (ALB), 41 (TRF / ALB), 42 (PAB), we must provide 6x3x2x2x2, i.e. 144 additional rules But the high TRF must be interpreted according to sex and age. However, out of 144 rules, a high TRF is observed 48 times. 48 x2 (sex) x 3 (age) must therefore be provided, i.e. 288 additional rules. Moreover, within sex and age, the criterion of menopausal woman or not intervenes, which leads to 1 2x6, or 72 additional rules.

The total number of rules is therefore 144 + 288 + 72, or 504 rules. But, in fact, 96 rules prove to be impossible. There are therefore 408 different possible rules. The 6 resulting conclusions are: 1) CINF301: no inflammatory or very slight reaction:

2) CINF302: inflammatory reaction based solely on the increase in CRP

3) CINF303: inflammatory reaction with single protein increase in chronic reaction proteins

4) CIIMF304: frank inflammatory reaction with normal CRP 5) CINF305: frank inflammatory reaction with increased CRP

6) CINF306: lowering proteins of chronic inflammation

The 408 different possible rules include:

RINF307 = CINF301 + 139.1 + 140.1 + 141 .1 142.1 = CINF307

RINF738 = CINF306 + 139.3 + 140.2 + 141 .2 + 142.2 = CINF738

We will now consider the interpretation of Ig (Immunoglobulins) in the protein profile. This concerns items 131 (IgM), 132 (IgG) and I33 (IgA). The interpretation is different if there is a monoclonal protein or not. However, the presence of a monoclonal protein is not seen on the protein profile but on another analysis which is the electrophoresis of proteins.

However, this electrophoresis is not always required with a profile. Also, if you run it, you rarely find a monoclonal protein. When a monoclonal protein is found, the interpretation stops there, and this conclusion is not linked to a reaction inflammatory. In fact, the interpretation of the Ig begins with the processing of item 44 “Protein electrophoresis”. The answer can be:

- yes with the presence of a monoclonal protein (1 finding ^era) - yes with no monoclonal protein (new ^1st conclusion),

- no (new ^1st conclusion).

Any individual can present Ig levels outside the norms without this being pathological. What is pathological is the variation in this Ig rate on two samples, hence the treatment of item 7 "prior art".

If the answer is no, this results in a 2 ^{nd β} general conclusion before the actual processing of Ig.

Then link items 31, 32, 33 to the inflammatory reaction. The conclusions of the interpretation of Immunoglobulins were reduced to 5 according to an approach similar to that adopted for inflammatory reactions: no mild inflammatory reaction inflammatory reaction - inflammatory reaction only due to CRP inflammatory reaction present (1, 2 or 3 proteins )

- decrease in proteins of the inflammatory reaction. The ^3rd conclusion is therefore selected from the following rules:

131 x I32 x I33 x C5 5 x 3 x 5 x 5 = 375 new rules.

The first conclusion can be established as follows: R | G1 = 1 44.2.1 + 12.1 = CIG 1

RIG13 = I7.2 = CIG1 3

CIG1 = Electrophoresis and immunoelectrophoresis have demonstrated monoclonal IgM. Given the age of the patient, we must first think to a severe subacute or chronic infection of viral or bacterial origin. This suggests an associated immune deficiency.

CIG1 3 = The Ig values reflect all the defenses acquired during life as a function of encounters with the various pathogens. In adulthood, in a healthy person, this rate varies little. It is therefore quite possible that a rate outside the norms has no pathological connotation, but is a completely physiological rate for the patient. What is interesting is the appreciation of the variation over two direct debits several months apart. Having no prior experience for this patient, the different etiologies proposed are only for information because interpretation must be done above all according to the clinical context.

As indicated above, the 60 different conclusions of the inflammatory reaction are reduced to 5, so that the following conclusions are determined: No inflammatory reaction CINF1 or CINF2 or CINF3 or CINF1 7 or CINF21 or CINF22 or CINF23 or CINF6 or CINF1 6 or CINF1 8 or CINF26 or CINF76 or CINF77 or CINF78 = CIG101 Mild inflammatory reaction

CINF7 or CINF8 or CINF31 or CINF83 = CIG102 Inflammatory reaction due only to CRP

CINF4 or CINF5 or CINF1 9 or or CINF80 = CIG103

Inflammatory reaction present (due to one or more proteins) CINF9 or CINF10 or ...... or CINF34 ... or CINF90 = CIG104

Decrease in proteins CINF91 or CINF92 or CINF93 or CING94 or CINF95 or CINF96 or CINF97 or CINF98 or CINF99 or CINF100 = CIG105 Then establishes the rules 375 for the ^3rd conclusion, as for example:

RIG 106 = CIG 101 + 131 .1 + 132.1 + 133.1 = CIG106

RIG548 = CIG105 + 131 .5 + I32.3 + I33.5 = CIG548

Additional rules based on age are added to take into account situations where, each time there is an inflammatory reaction (CIG104) without any increase in Ig, or with a decrease in IgM. To create these additional rules, some of the aforementioned CIGxxx conclusions are grouped together to lead to 5 conclusions CIG1 200, CIG1 201, CIG1 202, CIG1 203, CIG1 204 which are used in the establishment of these additional rules. We will now expose an example of implementation of the data processing method according to the invention for a combined interpretation of the genetic profile and the cardiovascular risk.

First, a non-exhaustive list of genes that can be interpreted in the context of the expert biological analysis system according to the invention is provided in Table I below. For each gene, a symbol + in a column means that this gene intervenes in the characteristic corresponding to this column, and conversely a symbol - in another column means that the same gene does not intervene for the characteristic corresponding to this column other column. Thus, by way of example, the CYP1 A1 gene intervenes in the smoker and in matters of nutrigenetics, but not in matters of pharmacogenetics, in immunogenetics and for oxidative stress. Thus, for each + symbol in this table, there are links and rules that must be written and integrated into the expert system. The following are given, by way of nonlimiting example, extracts of biological interpretation provided by an expert system according to the invention, with regard to the cardiovascular risk: “In view of the biological results, there is no atherogenic risk. It is therefore necessary to explore the other risk factors because nearly 20% of patients who have cardio-vascular problems have normal or subnormal biology. "[...]

“Hyperhomocysteinemia due to a congenital deficit in the enzymes involved in its biosynthesis is much rarer. For example, the cystathionine beta-synthetase deficiency is estimated at 1/20000 subjects who, in addition to the cardiovascular risk, also have mental retardation, dislocation of the lens, bone deformities. On the other hand, the deficit in 5-10 methyline tetrahydrofolate reductase is more frequent because estimated at 5% of the general population and is the major cause of genetic predisposition to a moderate hyperhomocysteinemia. These patients often have cardiovascular disorders in the first years of life. ...] »This is an indication to carry out genetic tests to find out if the increase in homocysteine is of genetic origin or not. "If the E2 allele seems to play a role in type III hyperlipoproteinemias, the E4 alele is also more involved in cardiovascular diseases. The E2 / E4 genotype, although uncommon, therefore significantly increases the risk of cardiovascular problems. In general, the average cholesterolemia of E4 / E3 subjects is higher than that of E3 / E3 subjects, itself higher than that of E3 / E2 subjects. Likewise, the average LDL cholesterol concentration of E4 / E3 subjects is higher than that of E3 / E3 subjects, itself higher than that of E2 / E2 subjects. On the other hand, the triglycerides are significantly higher in E2 / E2 subjects; E3 / E2; E4 / E2 than in subjects E3 / E3; E4 / E3 " This conclusion reflects a direct relationship between interpretation of a genetic profile and interpretation of a biological profile (cholesterol, triglycerides).

Below is an example of a conclusion reflecting a direct relationship between genetics and diet:

“People with E4 al / è / e are more sensitive to lipid-lowering and cholesterol-lowering diets. In these same subjects, the return to a diet rich in fats, in particular saturated fatty acids, leads to a greater increase in plasma cholesterol. "

The expert system according to the invention can also take into account in the conclusions supplied to the user a direct relationship between the interpretation of the genetic profile and data on the drug treatment obtained from personalized information on the patient, such as the 'illustrates the conclusion presented below:

“Subjects with f2 / e / e2 and suffering from hyperlipoproteinemia type lib respond well to treatment with gemfibrosil and with statins (simvastatin and lovastatin). Among subjects with type III hyperlipoproteinemia, carriers of the E2 or E3 allele respond well to treatment with statins.

On the other hand, subjects carrying the E4 allele would respond less well to lipid-lowering drug treatments, with the possible exception of probucol. "

Of course, the invention is not limited to the examples which have just been described and numerous modifications can be made to these examples without departing from the scope of the invention. In particular, provision can be made for complete automation of the operations for determining the biological profile and the genetic profile of a patient and for the combined treatment of these profiles. Furthermore, it will be readily understood that an expert biological analysis system according to the invention can also be coupled with databases and knowledge bases. Moreover, in the context of the present invention, the biological profile includes not only several families of biological determinations and analyzes now well established such as the protein profile or lymphocyte typing, but also other profiles being developed or which will be proposed in the future. Similarly, the expert system according to the invention aims to take into account increasingly complex genetic profiles as scientific and technological advances in this field.

Table I

Pharmacogenetics Immunogenetics Smoker Stress O. Nutrigenetics

+ + + +

+ +

+ + + +

+

Table I (continued)

Claims

1. Method for processing biological analysis data comprising processing a set of data relating to the biological profile of a human or animal subject, in the form of biological items to which a set of rules are applied providing conclusions in the form statements in natural language, characterized in that said set of rules includes a plurality of groups of rules each associated with a group of analyzes from among a plurality of groups of analyzes constituting the biological profile, and in that it comprises in addition at least one operation of grouping together a set of conclusions from a group of rules, so as to restrict the number of conclusions which are entered into a following group of rules.

2. Method according to claim 1, characterized in that it further comprises a processing of genetic data relating to this patient in the form of a set of genetic items associated with a set of genes studied in this patient, this processing including the execution of rules of genetic interpretation applied to both biological items and genetic items.

3. Expert biological analysis system, implementing the data processing method according to one of the preceding claims, this system comprising: - means for collecting data from biological measurements carried out on a human or animal subject and defining a biological profile, and personal data relating to said human or animal subject, these data being represented in the form of a set of items, - an engine comprising several groups of pre-established rules, each of said rules relating to certain items selected from among 'together of items and / or on certain conclusions of rules executed prior to said rule, and

- Means for issuing conclusions in response to the execution of said rules, characterized in that it further comprises means for grouping a set of conclusions from a group of rules associated with a group of analyzes among a plurality groups of analyzes constituting the biological profile, so as to limit the number of conclusions which are entered in a following group of rules.

4. Expert system according to claim 3, characterized in that it further comprises means for collecting data relating to the genetic profile of said human or animal subject in the form of genetic items, and in that the engine further comprises rules for the interpretation of said genetic items.

5. Expert system according to one of claims 3 or 4, characterized in that it is integrated into an automated system including operations for determining the biological profile and the genetic profile of patients.

6. Expert system according to one of claims 3 to 5, characterized in that it is coupled to a knowledge base.

7. Expert system according to one of claims 3 to 6, wherein the biological profile includes a protein profile.

8. Expert system according to one of claims 3 to 7, wherein the biological profile includes a lymphocyte typing.