The health data journey, from device to doctor

The Internet of Medical Things brings the sensor-driven data collection and analysis that IoT has made so popular for consumers and industry, and brings it to healthcare to deliver meaningful insights for patient analysis and health, with the security and confidentiality that it requires.

Around the world, we bought over 100 million wearable devices like smartwatches and fitness trackers in 2018. Increasingly, those are used for monitoring health, with AI used both to respond to sensor readings immediately and also to analyse giant data sets in the cloud, in the hope of improving treatments and helping individual patients with personalised medicine.

The Internet of Medical Things (IoMT)

An increasing amount of medical equipment is now monitored with sensors that track how well the equipment is functioning but there’s also an emerging section of IoT designed for capturing health data – the Internet of Medical Things. Continuous patient monitoring with instrumented beds and infusion pumps or wearables like chest monitors, tracks a hospital patient’s condition and vital signs without a nurse visiting them several times a day, and staff get a notification when critical attention is needed. A cardiac patient might do better if they can get up and move around; a wearable cardiac monitor means they’re not stuck in a bed unable to move because of all the wires.

Doctors can also monitor patients outside the hospital using wearables, to reduce re-admission, track rehabilitation after an operation, or monitor long-term conditions like diabetes and heart disease, to see if patients are really taking their medicine and doing their exercises correctly or if they need more support. In hospital or at home, it’s more accurate and less intrusive to take those measurements with sensors.

Data collection and data security

The data gathered by these kinds of sensors is obviously sensitive; it’s protected health information covered by regulations like HIPAA, HITRUST and GDPR. There’s also a large volume of data to deal with that has to be collected in high volume with sub-second measurements. An IoMT solution might also take data from multiple sources with different devices tracking heart rate, blood pressure, glucose levels, and physical activity, responses from patients and even devices they might walk past. You also want a solution that can monitor the health of the sensors themselves, to make sure they stay accurate.

One way to get all of that is the combination of Azure IoT Central, IoT Hub and the open source IoMT FHIR Connector for Azure (which includes an FHIR HealthKit framework), together with either the Azure FHIR API as a Service or your own open source FHIR Server for Azure which you can spin up using Cosmos DB or SQL Server as a backend. The HL7 Fast Healthcare Interoperability Resources format is quickly becoming the approved standard for moving healthcare data around securely using REST APIs.

Azure IoT for heart health monitoring

Peerbridge Health uses Azure IoT for its wireless Cor ECG, which can record up to seven days of heart activity (even in the shower), send unusual measurements to the cardiologist in near-real time and let cardiac patients start exercising while being monitored for the kind of irregular heart activity that could indicate a clot or blockage. The company picked Azure IoT because it provides a secure data stream plus cloud tools that put the data into the formats doctors are comfortable working with.

The digilog (Digital and Analog Companions for an Aging Population) IoT prototype developed by a consortium of health organisations is even more ambitious, trying to collect ECG and blood pressure measurements from a fitness band and some tiny wearable sensors in real time using Azure IoT Hub, to detect heart problems (using Azure Machine Learning) and show them on a smartphone app built using Power BI dashboards – all before people notice symptoms and visit a doctor. The team picked Azure IoT because it can securely collect store and data from almost any device at low cost; as new sensors and devices arrive, that data can be integrated into the same system.

The garment as the computer

Sensoria Health is creating multiple healthcare solutions using Azure IoT Central and the Azure FHIR API, CTO and co-founder Maurizio Macagno explained to CodeMatters. “We aspire to create truly wearable systems and a system of doctors and patients who can take better care of their therapy at home with a full stack of technology that starts with fabric sensors we apply to what we wear – garments and footwear and accessories – and then adding electronics so we can acquire data from the sensors that we transmit via Bluetooth to scenario-specific applications. Those are analysed on a smartphone or watch and give some feedback to the user but we also send data back to the doctor at the end of the chain so the doctor can see information about their patient population and which patients they should focus on first.”

Diabetics often lose feeling in their feet so a small cut can quickly turn into an ulcer; left untreated that can mean amputation – which reduces their life expectancy as much as being diagnosed with cancer, Macagno warns: just three to five years. The best way to treat ulcers is a boot that stops pressure on that specific area of the foot, so it can be treated and heal, but you need to wear it all the time unless you’re showering or sleeping. Sensoria turned the Optima Molliter boot into a ‘smart boot’ by adding sensors that can detect patient activity level and, critically, if they’re removing the boot. That data goes to an app on their phone or smartwatch that uses signal processing and machine learning algorithms to understand what they’re doing and remind them to put the boot back on or give them positive feedback if they’re using it correctly.

Azure infrastructure for continuous data flow

The data is also sent to Azure and aggregated for each medical practice, so doctors get a colour-coded clinical dashboard where they can see which patients are using the boot and who they need to follow up with. “They might call a red patient immediately and have a conversation; tell them they’ve detected too much activity and they need to rest more or ask them why they’re not wearing the boot,” he explained. “That’s a conversation between the patient and the doctor that just couldn’t happen before because they didn’t have the data.”

The architecture uses multiple Azure services. “Using the Azure infrastructure we can send data continuously via IoT Hub and using IoT Central we can manage the logistics of the devices: which are assigned to which organisation, what’s the firmware status and do we need to push an update. With the IoMT Connector we extract the data from IoT Central into our FHIR Server, which is based on Cosmos DB.”

Sensoria can use the same architecture for multiple devices and scenarios, he said. “The flow of data is very much the same: we have some device at the edge that is directly associated with the patient, which has a dedicated algorithm specific to the scenario and there is a flow of data to the same infrastructure in the cloud that flows to the same HIPAA-compliant storage and we can expose it with a customised dashboard.”

End-to-end surgery experience

The next solution is an end-to-end system for custom knee replacement – using a 3D printed cobalt and titanium prosthesis – including monitoring patient rehabilitation after surgery using a knee band with the Sensoria Core sensors and an app that can determine their range of motion in their therapy exercises, and how well they get back to full mobility. As well as the quantified information from the sensors, the app asks them how they feel about the surgery and their therapy and that feeds into the clinical dashboard, so doctors can see how many patients are able to do their exercises, whether they’re in pain or not and how they’re progressing.

Over time, the data sets that Sensoria collects will be anonymised and used to improve the algorithms in the wearable devices or for further research, using Azure Machine Learning for scale. “There are certain things even doctors don’t know about how types of activities can promote or worsen the quality of therapy,” Macagno noted.

Maximising open source tools and Azure services

He’s keen to start using a DICOM server Microsoft is developing to complement the FHIR server, for storing medical images like CT scans. He also pointed out how helpful the mix of Azure services and open source tools is for helping ISVs and startups who might not have the resources to build tools they need, but can pick up an open source component like the IoMT Connector and extend it. “It’s a simple but versatile component that’s very customisable to the needs of a company that has to connect data streams with medical storage. Because it’s open source you can contribute: if there’s something that your scenario requires that’s not there, you can take a stab at extending it, and if it’s useful it will most likely be incorporated back into the main branch. The direction Microsoft has taken towards open source is very beneficial for embracing these technologies.”

This blog post is the third in a series of articles about Microsoft technology in the healthcare market.

Read the first in the series, “What Microsoft offers the healthcare market” post here

Read the second in the series, “What Microsoft is doing in healthcare research” here

Grey Matter has many ISV and developer customers who are utilising technology for meaningful impact. Our team of Azure technologists and cloud architects can consult with you on your project and discuss your use case. Call direct on +44 (0)1364 654200 or email: [email protected]