"The most dangerous decision-making fallacy is that informed decision-makers will naturally make better, more objective decisions. Making consistently timely, effective, informed decisions takes hard work. Trust me – it’s worth it. Effective decision-making is the essential common ingredient behind every successful step, initiative and strategy that people, organizations and national governments undertake."Ozzie Paez
The emerging healthcare revolution would be impossible without cutting edge technologies that were unavailable just a decade ago. These include miniature sensors, wearable and portable connected devices, remotely accessible big data storage systems, and a growing family of analytical methods and machine learning software. Together they are providing increasingly sophisticated health monitoring and assessment capabilities that will change the art, practice and business of medicine. They will also help us live longer and better as we shift from "fixing and curing what ails us" to staying healthy longer than we thought possible just a generation ago. In this post I'll introduce these marvels of science and engineering, and illustrate their promise with a representative device, system and service that are already making a difference.
Sensors, portable and wearable technologies. Sensors are small devices designed to measure physical characteristics like temperature, chemical composition, respiration and muscular activity. Portable and wearable technologies integrate sensors with smart hardware and software to capture, process and transmit biological, positional and other measures about the user. Blood pressure monitors, heart rate monitors, glucose meters, accelerometers and GPS capable devices are examples of portable and wearable devices that incorporate a variety of sensors.
Data cloud. The cloud generally refers to high capacity data storage centers accessible through the Internet and related services. Big Data refers to large capacity storage for data of varying formats (nonhomogeneous), which are usually hosted in the cloud by dedicated vendors like IBM, Google, Microsoft and Amazon. These systems provide storage and access to massive quantities of data, which are indispensible to the application of data analystics and machine learning.
Data analytics and machine Learning. Data analytics is a multidisciplinary approach to uncovering insights, trends and meaning from sets of data. It usually involves numerical analysis including statistics, probability and other forms of quantitative calculations. Data analytics provide decision makers with information and insights that influence, shape and drive their decisions. The outputs of data analytics are frequently presented in on-line reports and dashboards updated in real-time as new information flow into the system.
Machine learning is software designed to identify patterns in data, which in turn trigger changes in the software that result in improved predictions of similar patterns in future data. In other words, the software ‘learns’ from existing data what it should expect to see in the future and adapts without active human involvement. Machine learning systems create models of the world based on information and mathematical algorithms, which get automatically updated as incoming data change. For example, machine learning systems identify patterns in e-mail traffic and human response, which are then used to classify the desirability, undesirability and importance of future e-mails. The system learns as users interact with their e-mails and adjusts its model to improve the accuracy of future e-mail classification. That’s how SPAM filters are automatically updated on most large e-mail systems such as Gmail.
Putting it all together – an example. Alivecor is the innovative company behind Kardia, a single channel, inexpensive electrocardiograph (EKG) device. Kardia was designed so that individual users can take their own EKGs, which are transmitted to the cloud and then processed to detect signs of atrial fibrillation, a condition known to cause strokes. The device is small enough to fit in a pocket or purse, or can be attached to the back of a smart cell phone. It uses two sensors to capture electrical signals generated by the beating heart. These are then wirelessly transmitted to the connected cell phone, which in turn uploads it to the cloud, where machine learning algorithms analyze the readings and report the results back to the user. The system warns the user when it detects the presence of atrial fibrillation, and urges him to contact a doctor. Alternatively, Alivecor offers a more detailed evaluation for a small fee by a cardiac technician or Board-Certified Cardiologist. A technician's review is completed within one hour, while the more detailed cardiologist evaluation takes twenty-four hours or less.
Alivecor offers the Kardia device and related services that allow users to self-monitor for signs of atrial fibrillation, a known stroke risk factor.
I became intrigued by Alivecor and Kardia several years ago and started using their system. It initially tagged many of my readings as unclassified, but its machine learning algorithms have predictably improved over time, and I now get more consistent feedback that my heart rhythm is normal. Kardia’s convenience and low cost contrast sharply with the alternative, which involves a time consuming and relatively costly doctor’s visit. More importantly, the simple device and smart phone application allow users to self-monitor and contact their physician or cardiac specialist at the first sign of trouble, rather than risking a stroke by waiting for symptoms to appear months or years later. These innovative capabilities would have been unthinkable before the advent of small sensors, the data cloud and machine learning technologies – and they're just the beginning!
The next post will focus on the devices and systems that measure and peer into our bodies without invasive surgery, and discuss how new technologies are extending these capabilities beyond medical settings to our daily lives.