ML is all the buzz these days. ML has the potential to provide game-changing advantages to early adopters. Proving the concept that ML works for your business and that it can be used in your day-to-day business are entirely different. Powerful public-domain and proprietary platforms have emerged to allow enterprise-grade deployment of ML applications. A team of professionals experienced in ML is critical to success.
Patients are at the center of a life-sciences company’s supply chain. Millions of patients are diabetic but are not aware of it and hence are at risk for heart diseases. EMRs capture patient data and are gold mines of useful information. Machine Learning can sift through millions of patient records and learn to diagnose diabetes just by using EMR data. This helps to not only identify diabetic patients but also predict diabetes in patients as early as 12 months – all by using EMR data. This is a powerful tool for physicians and hospitals.
Today’s methods for achieving material analyses in a large supply chain are cumbersome as they rely on large analytical laboratories and costly sensors. Near infrared (NIR) sensors allow for quick and cheap generation of spectral information for materials. Using ML, the accurate learning of complex spectra data from NIR sensors can be achieved which can dramatically bring the supply chain costs down. The benefits are in the form of reduced waste, increased quality, reduced product recalls, preserving brand value via identification of counterfeit products, etc.
This research was done in collaboration with a large life science company under an R&D grant from the National Science Foundation (NSF). We demonstrated that Type II diabetes can be predicted by using machine learning and EMR data. The cost of false-positives is not significant in Diabetes while the cost of false-negatives is very high. Physicians can be proactive in their advice to patients using our model.