Research and Exploration



Biosensors, IoT and machine learning in healthcare

David White

David White

Market Research Analyst,

Philips Connected Sensing


We teeter on the brink of the next revolution in Healthcare Information Technology. Three concepts – wearable biosensors, the Internet of Things (IoT), and machine learning – will take us over the edge.  

Devices and sensors connected to the IoT continues to rapidly expand


IoT, and machine learning are becoming pervasive – even inevitable - in many industries.  Synergistically, they feed on each other. Each technology adds value to – and drives the need for - the other. The Internet of Things is being driven by the emergence of low cost sensors, communication, and processing. Mobile phones were one of the first devices participating in the Internet of Things – often, I suspect, without some owners realizing how big a part their phones played… 


Here are two example apps: weather forecasting and navigation. One popular weather app has been downloaded by tens of millions of smartphone users. Many of those users – perhaps unwittingly - have given permission for the app author to use their phones to sample barometric pressure and send that data back to the cloud for aggregation and analysis. Essentially, their phones are part of a massive, IoT-based, weather station. Each phone provides precise localized data that together can be used to improve the accuracy and granularity of weather forecasts overall.


Likewise, smartphone-based road navigation apps work in a similar way. During operation, apps typically provide frequent updates, suggesting better routes to users in real-time as traffic conditions change. But, how does the app know where congestion is building up, and where traffic is flowing freely? Because the phones of people using the app provide data on exactly where they, and how fast they are moving using the built-in GPS sensor. This doesn’t work well if only a few people use the app. But, it is estimated that at least 10% of the population of Los Angeles use one popular app1. Navigation apps are, from one perspective at least, IoT applications.  The apps use data from millions of devices to help alleviate some of the misery that commuters face by providing faster routes as they open up.

With both examples above, ample data is generated to feed the algorithms that drive the app.  That data is also going to be used for analytics.  For example, navigation apps may be free to use, but supported by advertising.  Drive along a certain highway, at a particular time of day, and an ad might pop-up for a nearby restaurant.  For that advertising model to work, potential advertisers need to gain insight into exactly how the application is used:  How many people are travelling where, at what time, on what days.  That insight comes from analytics – skilled analysts who pour over the data to gain insights that can give them a competitive edge.  Here’s an example, showing how this data was used to discern the dining habits of particular cities over Thanksgiving2.

80 percent consumers
We’ll need help from smart software, such as machine learning or artificial intelligence, to bridge the gap between what data we can assimilate, digest and interpret with human ingenuity alone”

Internet of things 01
By 2020, 40% of IoT-related technology will be health-related, more than any other category, making up a $117 billion market9
Internet of things 03
Telemedicine, telehealth, and M-health markets are anticipated to reach $45.4 billion by 2021.10
Internet of things 05
By 2020 each person will own an average of 7 connected devices.11

From IoT to Machine Learning


But, there’s a problem. As we accumulate more data, faster, from a growing range of IoT sources, we get what computer scientists call big data. And, once we have a big data problem, conventional approaches to analytics may be rendered impotent. Why? Because there will be too much data for us to assimilate, digest and interpret with human ingenuity alone. In short, it becomes increasingly difficult for mere humans to sort the wheat from the chaff. We’ll need help from smart software, such as machine learning or artificial intelligence, to bridge that gap. Machine learning can help to spot patterns and trends that are invisible to the naked eye because of the overwhelming amount and complexity of data. Consequently, machine learning algorithms can be very adept at finding solutions to problems that can’t easily be quantified or codified by rules. Such as identifying pictures of cats3


Beyond the frivolous feline applications of machine learning, how does machine learning fit into healthcare? More specifically, how do machine learning and IoT work in tandem to help both clinicians and patients? Currently, funding of artificial intelligence startups in healthcare is at record levels4, with applications in many areas being explored5. But, as the amount and complexity of healthcare data grows – fed at least partly by wearable biosensors, so will the potential for machine learning.  Ultimately, wearable biosensors are IoT devices. A wearable biosensor may generate a continuous stream of data – about a person’s vital signs, for example.  This could happen in any setting – acute care hospitals, skilled nursing facilities, or in the home for chronic conditions. That data may be used for simple monitoring and alerting6. Beyond that, machine learning may provide a raft of insights, finding patterns in data that lead to breakthroughs in therapies or treatment. Researchers are already exploring the use of machine learning to assist radiologists7. And a recent paper8 highlights the potential for machine learning to help health professionals establish a prognosis and improve diagnostic accuracy. All potentially from data accumulated from wearable biosensors.

Request more information and sign up to receive the latest news in connected care

* Required fields

1 At least 10% of Los Angeles is using Waze

2 Holiday Insights Series: Does Your City Like to Eat In or Go Out?

3 Google computer works out how to spot cats, BBC

4 Smarter Health: Record Deals To Healthcare-Related AI Startups, CB Insights

5 90+ Startups Transforming Healthcare with AI, CB Insights

6 Smart wearable body sensors for patient self-assessment and monitoring, NCBI

7 Enlisting artificial intelligence to assist radiologists, Stanford University School of Medicine

8 Predicting the Future — Big Data, Machine Learning, and Clinical Medicine, New England Journal of Medicine


10 Push Telecommunications for Tele-Medicine (PTT) and M-Health: Market Shares, Strategies, and Forecasts, Worldwide, 2015 to 2021

11 Cisco: “The Internet of Things: How the Next Evolution of the Internet is Changing Everythiing”, Aptil 2011

12 The Wearable Future, PWC, October 2014