Using advanced data analytics to predict rheumatoid arthritis outcomes

A recent study by researchers at Osaka Dental University, Kyoto University, Osaka Metropolitan University, and Osaka University of Electronics and Communications used advanced computing technology to analyze complex state transitions in patients with rheumatoid arthritis receiving drug treatment. The study, led by Professor Keiichi Yamamoto, published in the journal PLOS ONE, highlights the challenges faced by patients with rheumatoid arthritis in achieving stable remission, while proposing new ways to predict and improve treatment outcomes.
Rheumatoid arthritis is a chronic autoimmune disease characterized by inflammation of the joints, causing pain and disability. Despite advances in treatment, including the use of methotrexate and biological and synthetic disease-modifying antirheumatic drugs, only about half of patients achieve remission. This has led to the identification of a subset of patients classified as “difficult to treat” who do not respond adequately to conventional therapies. The main goal of the study is to better understand the stability of patient states over time and the response of these states to treatment.
The researchers performed energy landscape analysis and time series clustering on data from the Kyoto University Rheumatoid Arthritis Management Consortium cohort, which contains comprehensive clinical data from thousands of patients with rheumatoid arthritis. Energy landscape analysis, a method originally used in protein folding studies, was adapted here to assess the stability of states in patients with rheumatoid arthritis. By assigning energy values to different patient states, researchers can visualize and quantify how easily patients transition between stable and unstable states.
“Our research divides patient state transitions into two different patterns: ‘good stability leads to remission’ and ‘poor stability leads to treatment dead ends’,” Professor Yamamoto explained. The analysis showed that a large proportion of patients experienced status transitions that could be affected by treatment, but only patients in the “good stability” group consistently achieved remission. Energy landscapes provide clear visualization of which patients are likely to respond positively to treatment and which patients are not.
Time series clustering uses a method called dynamic time warping to further classify patients into three categories based on their state transitions over time: “Toward good stability,” “Towards poor stability,” and “Unstable.” Patients in unstable clusters present a particularly challenging situation because their clinical course is difficult to predict. “Patients in unstable clusters should be treated more carefully because their response to treatment is less predictable,” Professor Yamamoto emphasized.
The study also looked at the effects of different treatment strategies over three years, focusing specifically on the first six months of treatment, a critical window for achieving remission. The findings showed that most patients who eventually achieved remission showed significant improvement within the first six months, while those who did not improve during this period were less likely to do so later.
These insights into the dynamics of rheumatoid arthritis treatment emphasize the importance of early intervention and careful monitoring. The ability to predict which patients will respond to treatment could significantly improve outcomes by allowing for more personalized treatment planning. The study’s innovative use of energy landscape analysis and time series clustering provides clinicians with powerful tools to assess patient stability and make more informed decisions about treatment strategies.
The study concluded that energy landscape analysis is particularly useful in real-world clinical practice, where a patient’s condition changes over time and treatments need to be adjusted dynamically. This approach, combined with time series clustering, offers a promising way to address the complexities of rheumatoid arthritis treatment, particularly for patients who do not respond to conventional therapies.
As Professor Yamamoto said: “This study opens up new ways to understand how patients respond to rheumatoid arthritis treatments and may lead to more effective and personalized care strategies in the future.”
Journal reference
Yamamoto K., Sakaguchi M., Onishi A., Yokoyama S., Matsui Y., Yamamoto W., Onizawa H., Fujii T., Murata K., Tanaka M., Hashimoto M. and Matsuda S. ( 2024). “Energy landscape analysis and time series cluster analysis of multistability of patient states associated with drug therapy in rheumatoid arthritis: the KURAMA cohort study.” PLoS One, 19(5), e0302308. DOI: https://doi.org/10.1371/journal.pone.0302308
About the author
Dr. Keiichi Yamamoto Engaged in research and education in health data science and clinical research informatics, he has rich experience and good clinical research records in the establishment of numerous medical research databases. At Osaka Dental University, he is affiliated with the Department of Data Science, Industrial Research and Innovation Center, and Translational Research Institute, where he oversees investigator-initiated clinical trials for the development of drugs and medical devices. In addition, he serves as Director of the Education Information Center, managing IT operations for the entire university, including the hospital. His academic contributions include serving on the database management committees of several academic societies, as executive director of operations for the Society for Health Data Sciences, and as a board member of the Personal Health Records (PHR) Council.

Dr. Masahiko Sakaguchi Currently, he is an associate professor in the Department of Engineering Informatics, Osaka University of Electric and Communications, Japan. His research interests focus on applying operations research methods to health data. He is interested in analytical technologies that support decision-making by healthcare professionals. In addition, he is involved in the management of cancer registration databases and serves as a committee member of the Japanese Cancer Registry Association.