A new study by researchers from the Regenstrief Institute, Indiana University and Purdue University presents their low-cost, scalable methodology for the early identification of individuals at risk of developing dementia. Although the condition remains incurable, there are a number of common risk factors that, if targeted and addressed, could potentially reduce the chance of developing dementia or slow the rate of cognitive decline.
Detection of dementia risk is important for appropriate care management and planning. We wanted to solve the problem of early identification of individuals likely to develop dementia with a solution that is both scalable and cost-effective for the healthcare system.
To do this, we use existing information – passive data – already in the patient’s medical notes, for what we call a zero-minute assessment, at less than a dollar cost. Decision-oriented content selection methodology is used to develop an individualized prediction of dementia risk or to demonstrate evidence of mild cognitive impairment.”
Malaz Boustani, MD, MPH., study senior author from the Regenstrief Institute and the IU School of Medicine
This technique uses machine learning to select a subset of phrases or sentences from the medical notes in the electronic health record (EHR), written by a doctor, nurse, social worker or other healthcare provider, that are relevant to the intended outcome during a certain period. observation period. Medical notes are stories in an EHR that describe the patient’s health in free text form.
Information selected for extraction from the medical notes to predict dementia risk may include: physician comments, patient comments, blood pressure or cholesterol levels over time, observations of mental status by a family member or a medication history – including prescriptions and over-the-counter medications as well as “natural” remedies and supplements.
Predicting dementia risk helps the patient, family, and caregivers access resources such as support groups and the Centers for Medicare and Medicaid GUIDE model program, which help people live in their homes longer. It could also prompt doctors to eliminate medications commonly taken by older adults but are known to negatively impact the brain, and to have patient conversations about over-the-counter medications with similar characteristics. Knowing the risk of dementia could lead doctors to consider recently FDA-approved amyloid-lowering therapies that alter the trajectory of Alzheimer’s disease.
“Our methodology combines supervised and unsupervised machine learning to extract sentences relevant to dementia from the vast amount of medical notes available for each patient,” said study co-author Zina Ben Miled, PhD, MS, a Regenstrief Institute. affiliated scholar and former faculty member of Purdue University in Indianapolis. “In addition to improving predictive accuracy, this allows the healthcare provider to quickly confirm cognitive impairment by reviewing the specific text used to drive risk assessment by our language model.”
“Researchers from the Regenstrief Institute and Indiana University have been pioneers in demonstrating the utility of electronic health records since the early 1970s. Considering the enormous amount of effort required by both physicians and patients to capture EHR data , the goal should be to strive to maximize clinical value from these data, even beyond their central role in medical care,” said co-author Paul Dexter, MD, of Regenstrief and IU School of Medicine “through machine learning methods to apply to to identify patients at high risk for dementia in the future, this study provides an excellent and innovative example of the clinical value achievable with EHRs. The early identification of dementia will become increasingly important, especially as new treatments become available .”
Although the ultimate beneficiaries of using the new technique are patients and healthcare providers, providing zero-minute assessment at less than a dollar has a clear benefit for primary care physicians who are overburdened and often do not have the time and training required necessary to administer specialized cognitive skills. testing.
The study authors’ five-year clinical trial of their risk prediction tool, conducted in Indianapolis and Miami, is in its final year. The lessons learned from this trial will enable them to increase the usefulness of the framework for predicting dementia risk in primary care. The researchers plan future work on merging medical notes with other information in electronic health records and environmental data.
“Dementia risk prediction using decision-oriented content selection from medical notes” was published in Computers in biology and medicine. This research is supported by National Institute on Aging grant R01AG069765 from the National Institutes of Health (PIs: Malaz Boustani, MD, MPH; Zina Ben Miled, PhD, and James Galvin, MD, MPH).
Source:
Magazine reference:
Li, S.et al. (2024). Risk prediction for dementia using decision-oriented content selection from medical notes. Computers in biology and medicine. doi.org/10.1016/j.compbiomed.2024.109144.