Computing a weighted jaccard index of electronic medical record for disease prediction

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease. Patient’s spinal cord, brainstem, or motor cortex of the cerebral motor cortex are gradually degenerated and lead to systemic muscle atrophy and weakness. Medicine therapies used for treating ALS mainly focus on symptomatic treatment and delaying deterioration. At present, no precise treatment can effectively cure, halt, or reverse the progression of ALS disease. Therefore, it is extremely important to predict high-risk populations of ALS candidates and prevent from the early stages. This research analyzes the historical medical records of ALS patients from the Taiwan National Health Insurance Research Database (NHIRD). Through analyzing comorbidity within a specific time interval, we proposed a novel index, weighted Jaccard index, for effective prediction analyses. The weighted Jaccard index and logistic regression model were applied to build an ALS patient prediction system. Based on comparing electronic medical records testing subject and know ALS patient, we can effectively detect potential ALS patients at early stages. Early and accurate detection can provide medical doctors to conduct precise inspection and appropriate treatment with efficient and effective guidelines for decision making.