IoT and the promise of predictive health
When one of his 18-month-old triplets was diagnosed with Type-1 diabetes, Michael Maniscalco, CEO of Better Living Technologies, knew his family needed help. If his little boy’s blood sugar level got too low, he risked severe health complications, including death. To guard against this, the child was given a continuous glucose monitor patch that checked his levels and sent them to a smart phone, which then relayed the data to the cloud for assessment. If a reading dropped dangerously low, the family would be notified via the phone. But how could parents of toddler triplets monitor their phones 24/7?
They couldn’t. Within a month, both Michael and his partner were beyond sleep-deprived, and experiencing near constant stress. It wasn’t sustainable. Michael — who holds a computer science degree and had some experience with smart home devices — was convinced there had to be a better way, and set to work finding it. Ultimately, he connected his son’s glucose data in the cloud to any device in the home that could receive data: smart phones, yes, but also laptops, virtual assistants, smart watches, smart appliances and smart light bulbs. Now, no matter where they were or what they were doing, both parents could be confident they’d be alerted if something was wrong. And some semblance of sleep returned. By connecting cloud analytics to IoT devices, Michael had not only created a kind of personal health network for his son, but also a proof of concept for anyone needing remote 24/7 health monitoring.
As I listened to Michael’s story, it became clear to me that many of the IoT features people might use for personal healthcare are similar to those we’re creating today in smart, IoT-enabled factories. Especially in the realm of predictive analytics, using real-time statistics, data-mining, modeling, AI and machine learning to sift through data, find patterns, assess risk and then make predictions about the future. In industrial IoT, predictive analytics is the basis of tools, including predictive maintenance, real-time asset monitoring, quality sensing, supply synchronization and much more. Predictive analytics allows enterprises to be more forward-looking and proactive, anticipating human behaviors, infrastructure issues and outcomes based on data, not hunches or assumptions. Predictive analytics can even provide decision options.
Now, as we apply predictive analytics to health-related data, we begin to see the promise of predictive health. The glucose monitor Michael’s son uses takes a reading every 5 seconds and sends that information to the cloud. That’s 17,280 readings every day, 6,307,200 data points per year. Where a human doctor might be overwhelmed with this volume of data — or, more likely, just look at specific points in time — analytic tools can easily detect patterns within the full dataset. And that’s just for a single individual. In the future, anonymized data from every glucose monitor in the U.S. could yield truly massive datasets from which we extract patterns to design better treatments, and maybe even start predicting blood sugar problems before they arise. The same principle applies across all of health and wellness. Imagine what the data streams from fitness trackers and smart phones — like exercise, heart rate or sleep monitors — could contribute to long-term heart health.
The concept scales. Predictive analytics in healthcare has the potential to surface tremendous insights hiding in hospital data and in the massive data generated by wearable connected devices, like connected blood pressure monitors. Then there’s data soon to be collected by the hundreds of connected medical devices either new to market or on the way. As all these data begin to flow through analytics engines, the effects on chronic disease management and patient care will be extraordinary.
And it’s already happening. Medical centers are starting to use computer-aided detection systems — or computer-aided diagnosis systems — to help doctors interpret medical images like X-rays, MRIs and ultrasounds. Computer-aided detection systems combine artificial intelligence and computer vision with radiological and pathology image processing, and can be used to support preventive mammogram check-ups, colonoscopies and lung cancer imaging. The goal is to use technology to identify and detect the very earliest signs of abnormality in patients before the human professionals can, thus increasing chances of successful treatment. Predictive analysis techniques have already produced AI that is more accurate than radiologists and dermatologists in many areas of visual diagnosis, and this gap is expected to grow.
To be clear, I’m not suggesting that predictive analytics will replace the doctor and patient model of healthcare we’ve known for millennia. Rather, I’m talking about having machines do what they do best, so that healthcare professionals can do what they do best. IoT and predictive analytics can augment the doctor patient conversation and diagnoses with knowledge based on data — lots of data.
IoT in the healthcare market has more than doubled in the past five years, and shows no sign of slowing down. As more and more health data flows to the cloud for analysis, we’re rapidly approaching a time when healthcare professionals will routinely make computer-assisted diagnoses and treatment plans, and maybe even use predictive analytics to focus more attention on prevention. Unbeknownst to him, Michael Maniscalco’s son is already experiencing what the rest of his generation may one day take for granted: predictive health. It’s on the way.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.