U of M Seeks to License New Algorithm Predicting Mortality Risks for Seriously Ill Patients
The University of Minnesota is seeking to license a prototype algorithm recently developed by Medical School researchers capable of predicting one-year mortality risks for all kinds of seriously ill patients.
U officials say the aim of the algorithm is to cut down on unnecessary and invasive procedures which ultimately fail to enhance the quality of life of these frail patients nearing the end of their lives. Instead, by analyzing electronic medical record data, the algorithm can employ the “random forest model” to predict their risk of death within a year of the last day of hospitalization.
The hope is that, armed with an accurate mortality prediction, these seriously ill patients can be “empowered to make more informed health care choices” while at the same time enabling doctors to reduce unnecessary and stressful procedures on patients who are not likely to benefit from them – a practice which has been identified in patient surveys as a problem for hospitals.
The U earlier this month announced the publication of a proof-of-concept study for the algorithm in the Journal of General Internal Medicine co-authored by Dr. Nishant Sahni, an adjunct U of M Medical School assistant professor and the algorithm’s inventor. The study used data gathered from nearly 60,000 hospitalizations from six hospitals over four years to verify its accuracy.
It found the algorithm could reliably be used to estimate the risk of one-year mortality within a cohort of multi-condition hospitalized patients. It uses commonly obtained data from electronic medical records such as the last set of vital signs, blood counts, metabolic panels and demographic information (such as patient age and length of hospital stay) to estimate the mortality risk.
Meanwhile, the U’s Office for Technology Commercialization (OTC) is offering the algorithm for licensing opportunities.
Because of its potential to better allow healthcare providers to offer “appropriate and cost-effective care” to seriously ill patients, OTC says the technology can have applications as a “clinical decision support tool (i.e., for end-of-life care)” as well as for improving end-of-life planning and “risk adjustment for research.”
“Currently, there is no broadly applicable, well-validated model using commonly obtained labs and vitals from last-available hospital data to predict one-year mortality in a heterogeneous set of patients,” OTC’s marketing materials say.
The market opportunity for such a tool could be significant going forward due to the aging of the U.S. population — the number of Americans ages 65 and older is expected to reach more than 98 million by 2060, putting an increased strain on increasingly limited Medicare and Medicaid funding.
The algorithm could theoretically be used by hospitals as a way to improve their palliative and end-of-life care programs, allowing patients to use the information to exercise more control over the process.