Applying the same technologies used for voice recognition and credit card fraud detection to medical treatments could cut healthcare costs and improve patient outcomes by almost 50%, according to new research.
The research by Indiana University found that using patient data with machine-learning algorithms can drastically improve both the cost and quality of healthcare through simulation modeling.
The computer models simulated numerous alternative treatment paths out into the future and continually planned and replanned treatment as new information became available. In other words, it can "think like a doctor," according to the university.
This is not the first time artificial intelligence has been brought to bear on healthcare.
Last year, IBM announced that its Watson supercomputer would be used in evaluating evidence-based cancer treatment options for physicians, driving the decision-making process down to a matter of seconds. The Watson supercomputer was first offered to Cedars-Sinai's Samuel Oschin Comprehensive Cancer Institute in Los Angeles. Later that year, Watson was brought in to help Memorial Sloan-Kettering Cancer Center physicians diagnose and treat cancer patients.
The new research at Indiana University was non-disease-specific -- it could work for any diagnosis or disorder, simply by plugging in the relevant information. The research is aimed at addressing three issues related to healthcare in the United States: Rising costs expected to reach 30% of the gross domestic product by 2050; quality of care where patients receive the correct diagnosis and treatment less than half the time on a first visit; and a lag time of 13 to 17 years between research and practice in clinical care, the university said.
The research was performed by computer science assistant professor Kris Hauser and doctoral student Casey C. Bennett. The researchers used 500 randomly selected patients for the computer simulations.
The two researchers had access to clinical data, demographics and other information from 6,700 patients kept by the Centerstone Research Institute, a nonprofit provider of community-based behavioral healthcare. Between 60% and 70% of the patients had major clinical depression diagnoses but also had chronic physical disorders, including diabetes, hypertension and cardiovascular disease, which were used in the simulations.
Using real patient data, the researchers compared actual doctor performance and patient outcomes against computer decision-making models.
The artificial intelligence models obtained a 30% to 35% increase in positive patient outcomes, Bennett said.
"And we determined that tweaking certain model parameters could enhance the outcome advantage to about 50% more improvement at about half the cost, he said.
The cost of diagnosing and treating a patient was $189, compared to the treatment-as-usual cost of $497, Bennett said.
"The framework here easily outperforms the current treatment-as-usual, case-rate/fee-for-service models of healthcare," Bennet said.
The researchers used mathematical modeling frameworks, known as "The Markov Decision Processes" and "Dynamic Decision Networks" to perform the tests. The computer modeling considered all possible sequences of actions and effects of medical treatment in advance, "even in cases where we are unsure of the effects," Bennett said.
"Modeling lets us see more possibilities out to a further point, which is something that is hard for a doctor to do," Hauser added. "They just don't have all of that information available to them."
Previous work by Hauser and Bennett had shown how machine learning can determine the best treatment at a single point in time for an individual patient. This is the first time they used the computer modeling with a large group of patients.
Lucas Mearian covers storage, disaster recovery and business continuity, financial services infrastructure and health care IT for Computerworld. Follow Lucas on Twitter at @lucasmearian, or subscribe to Lucas's RSS feed . His email address is firstname.lastname@example.org.