I'm an Assistant Professor at The University of Tokyo, a Co-founder / Scientific Advisor at miup Inc., Technical Advisors at Medley, Inc. with expertise of Computational Statistics, Time-series Analysis (State Space Model, Data Assimilation), Applied Statistics, Bioinformatics, and Machine Learning. I'm interested in applying technological developments in computer science and biotechnology for the innovation of human being. In academics, I have two main topics. The first one is the development of sophisticated data assimilation methodologies, e.g., linear/nonlinear Kalman filters, in the field of computational and applided statistics. The second one is, as the main work in the current institute, human health data analysis such as time-series medical examination data and genome data, e.g., immunological cancer genome analysis, metagenome analysis, and biomedical researches for personalized and preventive medicine. In business field, the current interest is to develop several types of artificial intelligence systems utilizing, e.g., Bayesian statistical inference and natural image processing, for the application of medicine, health care, nursing care, and genetic engineering. I'm also studying modern history, classical strategy, e.g., Hundred Schools of Thought and The Prince, geopolitics and philosophy as life works.
Department of Intelligence Science and Technology
Graduate School of Informatics, Kyoto University
(Bioinformatics Center, Institute for Chemical Research, Kyoto University)
Thesis Title: Reconstructing Biological Systems Incorporating Multi-Source Biological Data via Data Assimilation Techniques
Department of Computer Science
Graduate School of Information Science, The University of Tokyo
(Human Genome Center, Institute of Medical Science, The University of Tokyo)
Thesis Title: Intracellular Systems Analysis by Assimilating Large Scale Biological Data and Pathway Simulation
Facalty of Science and Engineering, Waseda University
Baccalaureate Degree Program
Thesis Title: 遺伝子ネットワーク推定とデータ同化による細胞内ネットワークシミュレーションモデルの構築
We're researching human genome, transcriptome, epigenome and clinical data for personalized and preventive medicine through modeling, prediction and inference of disease and health systems. For example, we have partially focused on immunological cancer genome and metagenome analysis based on statistical science for these purposes.
We're trying to establish "Integrated Healthcare Eco-System" using IoT technology and clinical A.I. system for developing countries. Our system can triage and assess one's health risk without professional knowledge given symptoms and/or test values, suggest a candidate list of treatments, and support clinical decision for medical doctors. The system has a highly complicated statistical structure to enhance the quality of the inference, estimate the confidence level of the conclusions, and predict the future risks through Bayesian statistics and machine learning (We have no plan to publish the detailed statistical structures). If you're interested in the program, please send me a message.
We developed an automated consulting system termed Mogul for medical treatment. This system is a type of artificial intelligence and based on a simple Bayesian inference procedure (Its concept is similar to the one in miup Inc. but this version is a little bit older). Also, a pandemic prediction system for the degree of future influenza epidemic has been released in the website.
I'm directing to develop the system for making professional occupations more efficient, e.g., Administrative Lawyer and Shiho-shoshi, which are not enough creative to be done by human.
I directed to develop the system that can automatically consult nutritional advice through a mobile app and A.I. components.
Tohoku Medical Megaban Organization, Tohoku University
The main topic was genome wide association study and to develop novel strategies for analyzing rare variants association test. Our developed method is highly efficient and available to WGS data within a reasonable time. In addition, we proposed a novel statistical method termed kernel Bayes' approximate Bayesian computation filtering for estimation of parameter values in stochastic simulation models.
We developed automated evaluation systems for predicting the profit of books and cinema. The former system was integrated to Medley, Inc. as CRUNCH MAGAZINE. For the latter system, I still have advised the statistical perspective of applying the developed simulation system (CRUNCH CINEMA) to the collaborative companies. In addition, since we're collaborating with major studios and major entertainment companies in JAPAN and USA, we have extended and developed CRUNCH CINEMA for the application of general purpose.
Japan Society for The Promotion of Science
The research topic was Reconstructing Biological Systems Incorporating Multi-Source Biological Data via Data Assimilation Techniques.
[J1] T. Hasegawa, M. Nagasaki, R. Yamaguchi, S. Imoto and S. Miyano. An efficient method of exploring simulation models by assimilating literature and biological observational data, BioSystems, vol.121, pp.54-66, 2014.
[J2] T. Hasegawa, R. Yamaguchi, M. Nagasaki, S. Miyano and S. Imoto. Inference of gene regulatory networks incorporating multi-Source biological knowledge via a state space model with L1 regularization, PLoS ONE, vol.9(8), e105942, 2014.
[J3] T. Hasegawa, T. Mori, R. Yamaguchi, S. Imoto, S. Miyano and T. Akutsu. An efficient data assimilation schema for restoration and extension of gene regulatory networks using time-course observation data, Journal of Computational Biology, vol.21(11), pp. 785-798, 2014.
[J4] T. Hasegawa, T. Mori, R. Yamaguchi, T. Shimamura, S. Miyano, S. Imoto and T. Akutsu. Genomic data assimilation using a higher moment filtering technique for restoration of gene regulatory networks. BMC Systems Biology, vol.9(14), 2015.
[J5] T. Hasegawa, A. Niida, T. Mori, T. Shimamura, R. Yamaguchi, S. Miyano, T. Akutsu and S. Imoto. A likelihood-free filtering method via approximate Bayesian computation in evaluating biological simulation models. Computational Statistics and Data Analysis, vol.94, pp.63-74, 2016.
[J6] J. Kurashige, T. Hasegawa (Equal First), et al. Integrated molecular profiling of human gastric cancer identifies DDR2 as a potential regulator of peritoneal dissemination. Scientific Reports, vol.6, pp.22371, 2016.
[J7] E. Ayada, T. Hasegawa, A. Niida, S. Miyano and S. Imoto. Binary contingency table Method for analyzing gene mutation in cancer genome. International Journal of Bioinformatics Research and Applications (11th The International Symposium on Bioinformatics Research and Applications), vol.12(3), pp.211-226, 2016.
[J8] K. Kojima, Y. Kawai, N. Nariai, T. Mimori, T. Hasegawa and M. Nagasaki. Short tandem repeat number estimation from paired-end sequence reads by considering unobserved genealogy of multiple individuals. BMC Genomics (11th The International Symposium on Bioinformatics Research and Applications), vol.17(Suppl. 5), pp.465--476, 2016.
[J9] S. Koshiba, Y. Yamaguchi, K. Kojima, T. Hasegawa, et al. The structural origin of metabolic quantitative diversity. Scientific Reports, vol.6, pp.31463, 2016.
[J10] T. Hasegawa, K. Kojima, Y. Kawai, K. Misawa, T. Mimori and M. Nagasaki. AP-SKAT: highly-efficient genome-wide rare variant association test, BMC Genomics, vol.17(1), pp.1-8, 2016.
[C1] T. Hasegawa, R. Yamaguchi, M. Nagasaki, S. Imoto and S. Miyano. Poster: Comprehensive pharmacogenomic pathway screening by data assimilation, Proc. Computational Advances in Bio and Medical Sciences (ICCABS), 2011 IEEE 1st International Conference on, pp.246-246, 2011.
[C2] T. Hasegawa, R. Yamaguchi, M. Nagasaki, S. Imoto and S. Miyano. Comprehensive pharmacogenomic pathway screening by data assimilation, Proc. 7th International Symposium on Bioinformatics Research and Applications, volume 6674 of Lecture Notes in Computer Science, pp.160-171. Springer Berlin Heidelberg, 2011.
[C3] T. Hasegawa, S. Hayashi, E. Shimizu, S. Mizuno, R. Yamaguchi, S. Miyano, H. Nakagawa, and S. Imoto. An in silico automated pipeline to identify tumor specic neoantigens from whole genome and exome sequencing data. Proc. 12th International Symposium on Bioinformatics Research and Applications, 2016.
東京大学医科学研究所 ヘルスインテリジェンスセンター 助教。博士(情報学)。修士・博士課程に於いて情報科学を専攻、生命情報学研究（生体内システムの統計的解析手法開発）に携わる。2013年に情報科学技術と芸術の融合を行うべく、CRUNCHERS株式会社を共同創業し、自然言語処理技術を中核とした小説の評価システム(CRUNCH MAGAZINE)と統計的シミュレーションを用いた映画の興行収益予測システム(CRUNCH CINEMA)を開発(前者は2015年1月に事業譲渡)。京都大学大学院 情報学研究科博士課程を短縮修了後、東北大学東北メディカル・メガバンク機構助教に着任し、疾患と遺伝子変異の関連解析並びに希少変異の網羅的かつ高速な解析手法を開発を行う。また同時期からMedley, Inc.の技術顧問に着任し、ベイズ学習を用いた医療用疾患推定システム(Mogul)やインフルエンザの感染予測技術などを開発・提供する。現職では、複雑な性質を持つ時系列データの内部状態とその将来状態を正確に予測する統計理論研究、ICGCプロジェクトのメンバーとしての免疫ゲノムデータ解析、コホートスタディデータを用いた疾患の将来リスク予測モデル開発の他、症状や検査値から疾患の予知・予防を行う統計科学手法の開発・応用をmiup株式会社と共同で行っている。そのほか複数企業のAI(統計科学、機械学習)開発の技術顧問をする。生命情報学、統計科学、統計的シミュレーションが専門であり、特に状態空間モデルの理論的進展に関する幅広い知見を有している。
Research Bio: Takanori Hasegawa is currently an Assistant Professor of Health Intelligence Center, Institute of Medical Science, University of Tokyo. He received BS in Engineering from Waseda university, MS in Information Science and Technology from The university of Tokyo, and PhD in Informatics from Kyoto University in 2010, 2012 and 2015, respectively. His current research interests cover time series analysis using state space models, e.g., genomic data assimilation, Bayesian statistical inference to assess one's helath risks and help M.D. to diagnose one's diseases, immunological and cancer genome analysis.