Hitachi, Ltd., the University of Utah Health (U of U Health) and Regenstrief Institute, Inc. (Regenstrief) recently announced the development of artificial intelligence (AI) technology to enhance care for type 2 diabetes patients who require intensive treatment.
Type 2 diabetes affects one in every ten people, although only a tiny percentage of those with the disease require numerous medicines to keep blood glucose levels under control and prevent major consequences including eyesight loss and kidney problems.
Physicians may have inadequate clinical decision-making skills or evidence-based guidelines for selecting drug combinations for this smaller population of patients. To encourage the formation of general principles to guide decision-making, there is a need to increase the number of patients.
Combining patient data from numerous healthcare institutions, on the other hand, requires a high level of artificial intelligence (AI) competence as well as extensive experience constructing machine learning models with sensitive and complicated healthcare data.
Hitachi, U of U Health, and Regenstrief experts collaborated to create and test a new AI algorithm that examined electronic health record data from patients with type 2 diabetes in Utah and Indiana and discovered generalizable treatment patterns of individuals with comparable features. These patterns can now be utilized to aid in the selection of the best treatment regimen for a certain patient.
In the article “Predicting pharmacotherapeutic outcomes for type 2 diabetes: An evaluation of three approaches to leveraging electronic health record data from multiple sources,” some of the findings of this study were reported in the peer-reviewed medical journal Journal of Biomedical Informatics.
For several years, Hitachi had collaborated with U of U Health on the creation of a pharmacotherapy selection method for diabetes treatment. However, due to a lack of data, the algorithm was not always able to properly forecast more complicated and less common treatment patterns.
Moreover, combining data from several facilities proved difficult due to the need to account for changes in patient disease conditions and treatment medicines provided between facilities and regions. The project teamed with Regenstrief to improve the data it was working with in order to address these issues.
The new AI technology divides patients into groups with comparable disease conditions before analyzing their treatment patterns and clinical results. It then links the patient to the disease state categories and forecasts the patient’s range of possible outcomes based on various treatment choices.
The researchers tested how effectively the technique predicted favorable results based on treatment regimens provided to diabetic patients in Utah and Indiana. Even when two or more drugs were administered concurrently, the system was able to support medication choice for more than 83% of patients.
The study team hopes that in the future, it will be able to help people with diabetes who need complex treatment by evaluating the efficiency of different drug combinations and then deciding on a treatment plan that is best for them with the support of their doctors. Not only will this improve diabetes care, but it will also improve patient involvement, compliance and quality of life.
The three parties will continue to analyze and enhance the new AI method’s efficacy, as well as contribute to future patient care through more healthcare informatics research.
Hitachi’s efforts, including the actual use of this technology, will be accelerated by the partnership between its healthcare and IT business divisions, as well as its research and development department.
GlobalLogic Inc., a Hitachi Group company and a pioneer in digital engineering, is supporting healthcare-related initiatives in the United States and will expand collaboration in this area. The whole Hitachi group will contribute to the health and safety of the population as a result of these initiatives.
Tarumi, S., et al. (2022) Predicting pharmacotherapeutic outcomes for type 2 diabetes: An evaluation of three approaches to leveraging electronic health record data from multiple sources. Journal of Biomedical Informatics. doi.org/10.1016/j.jbi.2022.104001.