Method and system for determining a specific test to be performed in order to determine the biological condition of a biological subject

A system for providing computer-implemented support for analyzing biological systems and for determining an upcoming measurements procedure by determining a biological test to be performed on a biological subject in order to determine the biological condition of the biological subject, wherein the system comprises at least one processor and memory, the memory comprising instructions, which when executed by the at least one processor, cause the at least one processor to:

• create a basis representation s for a set of biological symptoms;
• read input data obtained, based on measurements, from a biological subject whose biological condition is to be determined, said input data comprising parameters quantifying a number of biological symptoms si, 1≤i≤N, associated with said biological subject, wherein the measurements enable parametrization of biological symptoms associated with the measurements;
• generate a representation s′ of said input data in said created basis representation s;
• apply a predictive model M(s) to said representation s′ of said input data in order to obtain a probability distribution d that yields the probabilities di that said parameters comprised in said input data corresponds to a specific biological condition yj, said predictive model M(s) being determined by applying a neural network on a set of data, said set of data comprising collected empirical data associating said biological symptoms s, with a determined biological condition yj;
• perform a first optimization of a probability di comprised in said probability distribution d with respect to a test indicating parameter x1, said test indicating parameter x1 being associated with a biological symptom sk different from said biological symptoms si, and being represented in said basis representation s, in order to obtain a first optimized probability value di*;
• perform at least one additional optimization of said obtained probability di comprised in said probability distribution d with respect to a different test indicating parameter x2, said test indicating parameter x2 being associated with yet another additional biological symptom sm different from said additional biological symptom sk and said biological symptom si, and being represented in said basis representation s, in order to obtain at least one additional optimized probability value di**;
• and
• determine a biological test to be conducted on said biological subject based on a comparison between said obtained optimized probability values di*, di**, and output data representative of said biological test to be conducted on the biological subject, thereby enabling a reduction of the number of conducted tests, which in turn reduces the strain on the biological subject and reduces time and costs associated with the analysis of the biological system.

A non-transitory computer-readable medium on which a computer program is stored, said computer program comprising instructions, which when executed by at least one processor, cause said at least one processor to:

• create a basis representation s for a set of biological symptoms;
• read input data obtained, based on measurements, from a biological subject whose biological condition is to be determined, the input data comprising parameters quantifying a number of biological symptoms si, 1≤i≤N, associated with the biological subject, wherein the measurements enable parametrization of biological symptoms associated with the measurements;
• generate a representation s′ of the input data in the created basis representation s;
• read a probability distribution d that yields the probabilities di that the parameters comprised in the input data corresponds to a specific biological condition yj, the probability distribution d being obtained by applying a predictive model M(s) to the representation s′ of the input data, the predictive model M(s) being determined by applying a neural network on a set of data, the set of data comprising collected empirical data associating the biological symptoms s, with a determined biological condition yj;
• perform a first optimization of a probability di comprised in the probability distribution d with respect to a test indicating parameter x1, the test indicating parameter x1 being associated with a biological symptom sk different from the biological symptoms si, and being represented in the basis representation s, in order to obtain a first optimized probability value di*;
• perform at least one additional optimization of the obtained probability di comprised in the probability distribution d with respect to a different test indicating parameter x2, the test indicating parameter x2 being associated with yet another additional biological symptom sm different from the additional biological symptom sk and the biological symptom si, and being represented in the basis representation s, in order to obtain at least one additional optimized probability value di**;
• and
• output a specific biological test to be conducted on the biological subject, the biological test being determined based on a comparison between the obtained optimized probability values di*, di**, thereby enabling a reduction of the number of conducted tests, which in turn reduces the strain on the biological subject and reduces time and costs associated with the analysis of the biological system.