Machine Learning Prognostic Models for Acute Leukemia and Decision Making Tools for Clinical Practice and Patients.

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K16975
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 52010:General internal medicine-related
• Research Institution: Niigata University
• Principal Investigator: Fuse Kyoko 新潟大学, 医歯学総合病院, 講師 (40783329)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 急性白血病 / 予後因子 / 機械学習

Outline of Final Research Achievements

We attempted to create a predictive model for the course of acute leukemia patients in order to synthesize multiple prognostic factors and customize treatment for each patient. First, we searched for novel prognostic factors to create the model. We identified the Marker chromosome as the molecular background of the disease, WT-1 mRNA expression thresholds, intensive oral care as supportive care to prevent complications, and HLA gene polymorphisms involved in tumor immune response.It was pointed out that the ADTree program has the potential to improve prognosis by changing the transplant source according to disease and stage prior to HSCT, and that preventive intervention may be possible by predicting the frequency of graft-versus-host disease.

Free Research Field

造血器腫瘍

Academic Significance and Societal Importance of the Research Achievements

機械学習による予後予測モデルで治療成功のための分岐点を抽出する可能性を明らかにできた。今後の個別化医療に応用が期待でいる。機械学習の精度を高めるためには、診療記録・データの質とばらつき、解析データの量などが今後の課題と考えられる。今後は、特に予備能が低い高齢者において治療選択と予後、医療費への影響に関する個別経過予測モデルの構築を試みたい。