Now let's turn to health care. In the health care space, there are a number of different kinds of businesses, ranging from hospitals to biotech, medical practices, rehab centers, senior living care, laboratory services, pharmaceutical, and medical device companies. These kinds of concerns revolve around developing new or expanded offerings, developing, testing, selling new drugs or medical devices, obviously improving care, optimizing resources, and predicting demand. Memorial Health Care System needed a transparent vendor verification process to help prevent fraud. They deployed an intelligent analysis platform that cross-references information among 800 different databases for greater visibility into vendor activities. The solution is generating over two million dollars per year in return on investment. Has reduced vendor invoice cycle times from one month down to 18 days on average, about a 40 percent time reduction. It's contributed a 70-million dollar cost reduction throughout the health care system. One of the issues with pharmaceutical drug promotion is what's called off-label usage. Pharmaceutical companies need to monitor whether their sales reps are promoting their drugs for this off-label usage. So KPMG worked with a company called Altruists, to combine multiple complex data sources such as from doctors, sales reps, prescriptions, sales rep expenses, and training attendance to generate risk ratings. They created risks score dashboard to help identify and understand which sales reps or doctors are at most risk in participating in this kind of behavior. This enables them to target appropriate outreach to these account reps, based on behavioral and data results. Here's an example of a solution that helps improve hospital-patient monitoring. It's a product called EarlySense, a pressure plate with piezoelectric sensors beneath the patient, that continuously monitor for respiration, pulse, and movement. The real-time alerting capability alerts nursing staff monitors and mobile devices. It also includes detailed graphic and tabular reports. It's a contact-free system that eliminates cuffs or leads and enables more patient freedom of movement. It's been able to reduce the length of stay by nine percent, reduced the number of ICU days by 47 percent, and reduce pressure ulcers or bedsores by 64 percent. Also, it's been able to reduce patient falls by 43 percent. Hospitals' report return on investment of up to two million dollars per year, for a 300-bed hospital. Another clever healthcare related analytic solution out of the University of Washington is called the BiliCam. It helps screen for jaundice in babies without having to visit the hospital. It's a smartphone application that checks for jaundice in newborns by taking a picture of the baby and a color calibration card at the same time. The data is then analyzed by a machine learning algorithm. It delivers results directly to parents and pediatricians within minutes. In a limited trial, it outperformed current medical apparatus that cost several thousands of dollars. At Mount Sinai Hospital, they created a system to discriminate between two commonly misdiagnosed diseases. Echocardiograms consisting of 10,000 attributes from 90 metrics in six different locations of the heart, all produced by a single one-second heartbeat. It involves some really sophisticated analytics including, an associative memory engine, that combines NoSQL, semantic graph, machine learning, cognitive-distance algorithms based on Kolmogorov complexities, and stuff like that. Anyway, the results are that it's able to discern between two completely different diagnoses, both of which cause heart failure, but are complex to diagnose, and require vastly different treatments. This system has been able to reduce the misdiagnoses from 27 percent down to 10 percent. A system developed by researchers and doctors at Boston Children's Hospital helps with the early identification of disease outbreaks. It automatically sifts through millions of posts on dozens of social media sites, local news reports, medical workers' social networks, and government websites to track instances of a disease. It uses sophisticated filtering algorithms to reduce the noise in that content. It continually plots disease hotspots on a map. One of the results was that it was able to identify a cluster of mystery hemorrhagic fever in Guinea over a week before the World Health Organization confirmed an Ebola outbreak. This kind of solution can really help organizations like the WHO, that are particularly cash-strapped. On average, it takes women four to six months to get pregnant. But a new algorithm developed by Harvard scientists and fertility experts can more accurately predict ovulation cycles, and when to plan romantic interludes. It uses four million data points from over 70,000 women via wearable quantified self device like a Fitbit, along with self-reported data on, body medical history, weight, menstrual cycles, and sexual intercourse specifics via a mobile app. The app now used by 40,000 women and their partners, has helped them become pregnant in about 60 days, or two to three times faster than the national average. Express Scripts wanted to identify and intervene with patients who are less likely to take their prescriptions correctly, like the elderly, especially patients taking meds for high blood pressure, diabetes, high cholesterol, asthma, osteoporosis, and multiple sclerosis. The system analyzes 400 variables including, prescription history, the economic makeup of the patient's neighborhood, and so forth. It's able to predict medication compliance, 90 percent of the time. The system uses tailored manual interventions, including phone calls, and help with signing up for auto refills. They've also introduced beeping bottle caps that increase compliance, two percent. Timers are given to forgetful patients to improve their compliance by 16 percent.
