First standalone version, FHIR generator create Device on http://hapi.fhir.org/baseR4/Device/1091456
Version 2, first version for testing with real buttons - leaving the devspace
Version 0.95 with LED, 8 configurations and status availability switch.
the button (still without LEDS, will add them later today)
The #WIRVSVIRUS hackaton topic collection showed that a big problem is currently ensuring that patients be rapidly getting the intensive care unit support they need. However, referring Physicians do not know where currently beds are available. This causes delays in provisioning care to the patient with subsequent higher letality rate. Additionally, it causes some hospitals to be overcrowded, while others with free capacities are unused. Furthermore, ICU resources are the highest cost for hospitals and therefore need to be utilised 100% and therefore overbooking (like in airplane industry) is a financial must.
What it does
Current approaches enables hospitals to be quickly enabled in tracking the status of the occupancy of their ICU and also of other resources that they want to share globally. It enables referring physicians to see where they can send their patients, and therefore enables addressing all the issues above. Additionally, it is also extendible to other uses beyond ICU beds. It does not require any changes to their hospital information systems and does not require any investment in iT. I
How I built it
The system consists of different components.
- an "internet button" that connnects to the local WIFI and sends a signal with status to the receiving server
- a server that receives notification of which resources are in which status (public FHIR HAPI server)
- a visualisation component that allows to show the status of resources ( it connects to the server that collect such information )
Challenges I ran into
Business / Hospital Processes / Regulatory in Germany
- Adoption of the approach amongst German hospitals might not be very high since the availabilty of beds could be considered a business relevant matter, even in emergency cases like the current one.
- Furthermore, with the lack of "key" to trigger the eButton, anyone could (voluntarily or not) trigger it - with the consequence of inappropriate information posted online.
- Finally, all hospitals in Germany have already an Hospital Information System, and this surely has amongst other features also the management of beds.
- The device itself would not be considered as a "medical device" however a certification might be required, therefore preventing a short term deployment. ### Technical
- Not all hospitals have yet a FHIR repository, and therefore the FHIR messages that the eButton sends need to be "translated" to format that hospitals are able to use, or custom code needs to be written.
- FHIR does not provide yet a Library for Arduino, I therefore started that too: https://github.com/lucatoldo/arduino-fhir
- https://arduinojson.org/v6/assistant/ is a great tool to generate code from JSON examples .
- Cryptography on ESP8266 is possible, and herewith demonstrated, however scarcely documented. The involvement of THE ESP8266 crypto expert (EFP3) was essential to quickly resolve this matter without additional hardware.
Accomplishments that I'm proud of
Quite some people found the idea good since it is simple and cheap and can "fly" even without touching or requiring existing IT landscape of the hospital. The problem of managing resource availability is a general one and there is need for IoT based decentralised solutions.
What I learned
- Grassroot approaches have the capability of freeing a lot of energy and come up with novel ideas, that might have been ignored by rigorous product design and market analysis. However, they risk of falling into the category of "looking for a problem" that does not exist, and lack of domain/business knowledge might result in wasted effort.
- Having an enthusiastic hardware designer (unixb0y) in the team is a must in order to do IOT work ! ### Technical
- ESP8266EX (NodeMCU, ESP-F) can be used as cryptographic pen, to digitally sign secure e-signatures.
- IOT is a great thing to address rapidly urgent needs with simple components.
- Digital Signature is a datatype of FHIR (https://www.hl7.org/fhir/datatypes.html#Signature) however it is not yet widely adopted.
- Building IOT devices that generate FHIR is possible but cumbersome, since there is no FHIR Library for Arduino.
What's next for ICU-bed-availability-service
If Hospitals or organisations will be interested in field test, I am willing to continue this project through various improvements, otherwise it will sleep in the depth of GITHUB .