Ethan Wood is a computer science student in his second year at UCL and as part of his course competed in the FHIR (fast healthcare interoperability resources) hackathon in the internet of things category building FHIR-Display. Ethan also developed FHIR-Parser, “an elegant and simple FHIR library for Python, built for human beings”.
One of the problems when viewing patient data inside of a hospital is the trade-off between privacy and ease of access. You need key patient information, such as their name or medication, to be readily accessible. Currently this is in a printed form, close to their bed, so that clinicians can easily monitor lots of patients at the same time. This allows anyone with physical access to a patient to be able to read their medical records. It can also be difficult to keep track of patients when they are moving around different parts of the hospital, especially if they are out-patients and not easily recognised. Further to this, more in-depth, up to date, or complex information is held digitally and needs to be accessed via a computer or iPad.
This presents two problems, the first is how do we improve patient privacy while ensuring it is just as easy for clinicians to access the data they need. The second stems from the inherent difficulty of using a multi-purpose device (such as a computer or iPad) which traditionally encompasses a variety of different tasks for a clinician. When using this multi-purpose device, they end up needing to switch between different apps and services, logging in and out. As FHIR is integrated into more of the hospital ecosystem this situation is only likely to worsen with yet another system they need to connect to being created.
This is solved by FHIR-Display, a dedicated device for clinicians to access FHIR data about a patient in a secure environment. The idea is that all patients during triage are issued with a small RFID tag which contains their personal FHIR identification code. This can be in the form of a bracelet, key fob, or security tag depending on what is most appropriate for this particular patient (in-patients may prefer a bracelet if it is frequently accessed while for out-patients a key fob they can remove). RFID tags are very cheap and there is no risk of data leaking outside as it only contains their FHIR identification code for that specific hospital. When a clinician needs to see data about a particular patient, they place the FHIR-Display near the RFID tag, at which point the patient’s personal information appears on the screen. This also gives patients a feeling of control about their personal data with them being able to prevent someone they do not trust from looking at it.
Currently FHIR display is a small box, approximately the size of a phone, built from 3D printed PLA. It houses an Arduino Mega for the controller, an LCD keypad shield allowing up to 32 characters to be displayed, and an RFID-RC522. The RFID tag is read using the RFID-RC522 through the PLA base which is designed to be thin enough for radio waves to penetrate. The device connects to the FHIR endpoint either over WIFI, Bluetooth, or serial. During the hackathon it connected to the FHIR endpoint via FHIR-Parser which aims to simplify the FHIR standard and allow projects to be built quickly without the developer needing to fully understand how FHIR works. This allows them to instead focus on the idea of their project and let the parser handle retrieval (ideal for short hackathons). However, in the future it will be possible to connect FHIR-Display directly to an endpoint such as the Azure API for FHIR with the ATmega2560 more than powerful enough to complete the parsing and interpretation independently.
While the idea presented above simplifies access to data and helps prevent “yet another app” on a clinician’s device, as of yet, it only improves security slightly by removing the need for bedside reports. However, anyone who has one of the devices is still able to access information about the patients. In the future this would be solved with some sort of biometric lock. This would most likely take the form of a fingerprint sensor with a registered list of users. Once the device detects that it has been pickup (through the inbuilt accelerometer) it will require authentication and stays active until placed back down.
Future work on the project will focus on increasing the amount of data which can be shown and reducing the footprint of the device. For the former using a small 3.5-inch TFT LCD touchscreen instead of an LCD keypad would allow the graphing of FHIR data and enable more complex actions, such as recording a patient’s blood oxygen levels), directly to the FHIR endpoint. During the prototyping stage, the footprint can be dramatically reduced by switching to an Arduino Nano and a more precisely printed 3D box. This could allow the device to become wearable and fit into other data display applications in the hospital. At production the device could be made smaller still, with a custom PCB and micro-controller.