Analyzing medical records from thousands of patients, statisticianshave devised a statistical model for predicting what other medicalproblems a patient might encounter. Like how Netflix recommends movies and TV shows or how Amazon.comsuggests products to buy, the algorithm makes predictions based onwhat a patient has already experienced as well as the experiencesof other patients showing a similar medical history. “This provides physicians with insights on what might be comingnext for a patient, based on experiences of other patients. It alsogives a predication that is interpretable by patients,” said TylerMcCormick, an assistant professor of statistics and sociology atthe University of Washington.
The algorithm will be published in an upcoming issue of the journal Annals of Applied Statistics . McCormick’s co-authors are Cynthia Rudin, Massachusetts Instituteof Technology, and David Madigan, Columbia University. McCormick said that this is one of the first times that this typeof predictive algorithm has been used in a medical setting. Whatdifferentiates his model from others, he said, is that it sharesinformation across patients who have similar health problems.
Thisallows for better predictions when details of a patient’s medicalhistory are sparse. For example, new patients might lack a lengthy file listingailments and drug prescriptions compiled from previous doctorvisits. The algorithm can compare the patient’s current healthcomplaints with other patients who have a more extensive medicalrecord that includes similar symptoms and the timing of when theyarise. Then the algorithm can point to what medical conditionsmight come next for the new patient. Glitter False Eyelashes
“We’re looking at each sequence of symptoms to try to predict therest of the sequence for a different patient,” McCormick said. If apatient has already had dyspepsia and epigastric pain, for instance, heartburn might be next. The algorithm can also accommodate situations where it’sstatistically difficult to predict a less common condition. Forinstance, most patients do not experience strokes , and accordingly most models could not predict one because theyonly factor in an individual patient’s medical history with astroke. But McCormick’s model mines medical histories of patientswho went on to have a stroke and uses that analysis to make astroke prediction. Diamond False Eyelashes Manufacturer
The statisticians used medical records obtained from a multiyearclinical drug trial involving tens of thousands of patients aged 40and older. The records included other demographic details, such asgender and ethnicity, as well as patients’ histories of medicalcomplaints and prescription medications. They found that of the 1,800 medical conditions in the dataset,most of them – 1,400 – occurred fewer than 10 times. McCormick andhis co-authors had to come up with a statistical way to notoverlook those 1,400 conditions, while alerting patients who mightactually experience those rarer conditions. They came up with a statistical modeling technique that is groundedin Bayesian methods, the backbone of many predictive algorithms.McCormick and his co-authors call their approach the HierarchicalAssociation Rule Model and are working toward making it availableto patients and doctors. Criss Cross Eyelashes Manufacturer
“We hope that this model will provide a more patient-centeredapproach to medical care and to improve patient experiences,”McCormick said. Additional References Citations.