Proactive Machine Learning Models Could Accelerate Healthcare Data Collection and Analysis
According to a viewpoint article published in JAMA.
According to Yuan Luo, PhD, associate professor of preventive medicine in the Division of Health and Biomedical Informatics and lead author of the publication.
Yuan Luo, PhD, associate professor of preventive medicine in the Division of Health and Biomedical Informatics, was lead author of the viewpoint published in JAMA.
“There is more research that needs to be completed before we can make this transformation, but it could apply to many areas,” said Luo, who is also director of AI at the Northwestern Clinical and Scientific Institute. Translational Sciences (NUCATS) and the Institute for Augmented Intelligence in Medicine, a professor at the McCormick School of Engineering and a member of the Robert H. Lurie Comprehensive Cancer Center at Northwestern University.
Machine learning (ML) algorithms are workflows that leverage statistical methods to glean useful patterns from data. In a conventional model, expert input is required at every stage, from data collection, to feature engineering and training, to eventual deployment and evaluation.
However, a proactive model automates some of these processes, such as assessment to inform new data collection, Luo said.
“It can close the loop from deployment to data collection, fine-tuning data collection without expert input,” Luo said.
A notable example of proactive ML was used in Greece, helping to target entry testing for travellers. The model, known as Eva, stratified travelers into risk categories based on age, gender and travel history: a more accurate estimation of risk than simply using the country of origin. This allowed Greek public health officials to allocate rare PCR tests to maximize detection of infected travellers.
The proactive part of this model was feedback; Eva automatically informed Greek officials that the risk estimates were based on little or old data and needed to be updated, helping to target new data collection that would eventually feed into the model.
“It told the human experts where they needed more data or if there was an emerging high-risk group, which can be helpful in a rapidly evolving pandemic,” Luo said.
Proactive ML models are still rare overall, likely due to their complexity and unfamiliarity, Luo said, but could be useful in areas with long timescales and many decision points, such as management. chronic care.
“It could help determine the treatment regimen, and each decision would impact the pattern and help inform the next decision,” Luo said. “If you can model the complex situation the patient is facing, that could be a game-changer.”
Ultimately, these models need to transition from virtual bench to bedside, Luo said.
“It is difficult for human experts to amass evidence and provide insights as we face increasing size and variety of data,” Luo said. “So we need to start deploying proactive ML systems in healthcare practice.”