# US 2018/0322958 A1 · Machine learning, natural language processing and network analysis-guided discovery related to medical research > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2018/0322958 A1 - Document ID: US20180322958A1 - Status: Published application - Kind: A1 - Application number: 15/966,454 - Filed: 2018-04-30 - Published: 2018-11-08 - Earliest claimed: 2017-05-04 - Inventors: Efthymios Kalafatis - Web page: https://patentgrep.com/patent/US20180322958A1 ## CPC - G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients - G06F 7/08: Sorting, i.e. grouping record carriers in numerical or other ordered sequence according to the classification of at least some of the information they carry (by merging two or more sets of carriers in ordered sequence G06F7/16) - G06F 16/24578: {using ranking} - G06F 16/2465: {Query processing support for facilitating data mining operations in structured databases} - G06N 3/0499: Feedforward networks - G06N 3/09: Supervised learning - G06N 5/041: {Abduction} - G06N 20/00: Machine learning - G06F 2207/4824: Neural networks - G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] - G06N 3/08: Learning methods - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 20/10: using kernel methods, e.g. support vector machines [SVM] - G06N 20/20: Ensemble learning - G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Techniques are disclosed for discovering, biological elements, functions and pathways, and environmental and nutritional factors related to diseases and medical syndromes. The techniques preprocess database query results and harmonize data using natural language processing before transforming them into the frequency space. The transformed results are analyzed with various categories of machine learning algorithms whose results are normalized, ranked and selectively combined, weighted or un-weighted, to produce a single result ranking the most important elements affecting a target disease or medical syndrome. The invention also uses alternative algorithms producing hypotheses on associations between medical topics, which are used as suggestions for exploratory medical research ## Claims ### Claim 1 A computer implemented method of discovering biological topics, symptoms and environmental and nutritional factors related to diseases, medical syndromes and biological functions, where biological topics comprise genes, biological molecules, and biological pathways, the computer implemented method comprising: querying a database for at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function; preprocessing the at least one query result for the at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function, where preprocessing comprises at least one of natural language processing, normalization, and natural language understanding; transforming the at least one preprocessed query result in a different feature space; applying a plurality of data mining algorithms to the at least one transformed result; post-processing the outcomes of the data mining algorithms; and selectively combining the outcomes of the data mining algorithms into a single result, where combining produces (a) biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions, and (b) patterns of interaction and effects of interaction between groups of biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions. ### Claim 2 (depends on 1) The computer implemented method of claim 1, further comprising: applying at least one algorithm to produce hypotheses on associations between (a) medical topics, symptoms, environmental and nutritional factors and (b) groups of medical topics, symptoms, environmental and nutritional factors with diseases, medical syndromes and biological functions. ### Claim 3 (depends on 1) The computer implemented method of claim 1, where preprocessing produces nodes used by a network analysis algorithm. ### Claim 4 (depends on 1) The computer implemented method of claim 1, where each of the data mining algorithms produces results of one of the following types: ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a significant confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a uniform confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values, where the ordering of the topics and the number of topics are different among algorithms; ordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions without any association confidence values, the ordered list comprising frequency of occurrence information; and unordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions. ### Claim 5 (depends on 4) The computer implemented method of claim 4, where combining the results of the data mining algorithms comprises normalization of the individual algorithm results for producing a ranking of the combined results, where the ranking of the combined results is different from the ranking of individual results and the ranking of the combined results is based on the confidence values produced by each algorithm. ### Claim 6 (depends on 1) The computer implemented method of claim 1, where a network analysis algorithm is used to identify important biological topics, symptoms and environmental and nutritional factors related to diseases, medical syndromes and biological functions by selecting those biological topics, symptoms and environmental and nutritional factors that have the highest score in at least one network analysis metric, where the at least one metric comprises centrality, degree, betweenness, and closeness. ### Claim 7 (depends on 6) The computer implemented method of claim 6, where the result of the network analysis algorithm is used to guide biological research by indicating biological topics for research. ### Claim 8 (depends on 5) The computer implemented method of claim 5, where the combined result is used to identify active drug substances having an effect on diseases, medical syndromes and biological functions. ### Claim 9 (depends on 5) The computer implemented method of claim 5, where the combined result is used to identify symptoms, groups of symptoms, and biological pathways associated with symptoms or groups of symptoms. ### Claim 10 (depends on 5) The computer implemented method of claim 5, where the combined result is used to verify experimental results and theories. ### Claim 11 A computing device configured to discover biological topics, symptoms and environmental and nutritional factors related to diseases, medical syndromes and biological functions, where biological topics comprise genes, biological molecules, and biological pathways, the computing device comprising a processor with means for: querying a database for at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function; preprocessing the at least one query result for the at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function, where preprocessing comprises at least one of natural language processing, normalization, and natural language understanding; transforming the at least one preprocessed query result in a different feature space; applying a plurality of data mining algorithms to the at least one transformed result; post-processing the outcomes of the data mining algorithms; and selectively combining the outcomes of the data mining algorithms into a single result, where combining produces (a) biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions, and (b) patterns of interaction and effects of interaction between groups of biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions. ### Claim 12 (depends on 11) The computing device of claim 11, further comprising: means for the processor to apply at least one algorithm to produce hypotheses on associations between (a) medical topics, symptoms, environmental and nutritional factors and (b) groups of medical topics, symptoms, environmental and nutritional factors with diseases, medical syndromes and biological functions. ### Claim 13 (depends on 11) The computing device of claim 11, where the processor has means to preprocess query results for producing nodes used by a network analysis algorithm. ### Claim 14 (depends on 11) The computing device of claim 11, where each of the data mining algorithms produces results of one of the following types: ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a significant confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a uniform confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values, where the ordering of the topics and the number of topics are different among algorithms; ordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions without any association confidence values, the ordered list comprising frequency of occurrence information; and unordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions. ### Claim 15 (depends on 14) The computing device of claim 14, where the means for combining the results of the data mining algorithms comprise normalization of the individual algorithm results for producing a ranking of the combined results, where the ranking of the combined results is different from the ranking of individual results and the ranking of the combined results is based on the confidence values produced by each algorithm. ### Claim 16 A non-transitory computer program product that causes a computing device to discover biological topics, symptoms and environmental and nutritional factors related to diseases, medical syndromes and biological functions, where biological topics comprise genes, biological molecules, and biological pathways, the non-transitory computer program product having instructions to: query a database for at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function; preprocess the at least one query result for the at least one biological topic, symptom, environmental or nutritional factor, disease, medical syndrome or biological function, where preprocessing comprises at least one of natural language processing, normalization, and natural language understanding; transform the at least one preprocessed query result in a different feature space; apply a plurality of data mining algorithms to the at least one transformed result; post-process the outcomes of the data mining algorithms; and selectively combine the outcomes of the data mining algorithms into a single result, where combining produces (a) biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions, and (b) patterns of interaction and effects of interaction between groups of biological topics, symptoms, environmental and nutritional factors associated with diseases, medical syndromes and biological functions. ### Claim 17 (depends on 16) The non-transitory computer program product of claim 16, further comprising: applying at least one algorithm to produce hypotheses on associations between (a) medical topics, symptoms, environmental and nutritional factors and (b) groups of medical topics, symptoms, environmental and nutritional factors with diseases, medical syndromes and biological functions. ### Claim 18 (depends on 16) The non-transitory computer program product of claim 16, where preprocessing produces nodes used by a network analysis algorithm. ### Claim 19 (depends on 16) The non-transitory computer program product of claim 16, where each of the data mining algorithms produces results of one of the following types: ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a significant confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values having a uniform confidence distance from each other; ordered list of biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions with association confidence values, where the ordering of the topics and the number of topics are different among algorithms; ordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions without any association confidence values, the ordered list comprising frequency of occurrence information; and unordered list of grouped biological topics, symptoms, environmental or nutritional factors associated with diseases, medical syndromes or biological functions. ### Claim 20 (depends on 19) The non-transitory computer program product of claim 19, where combining the results of the data mining algorithms comprises normalization of the individual algorithm results for producing a ranking of the combined results, where the ranking of the combined results is different from the ranking of individual results and the ranking of the combined results is based on the confidence values produced by each algorithm.