Authors

Inspiration

Disparate data sources and lack of interoperability in healthcare systems are issues we have been trying to address over the last couple of years. The current COVID-19 crisis has brought these issues into focus for everyone. The World Health Organization has defined a set of questionnaires that should be completed for each patient upon admission, transfer to an Intensive Care Unit (ICU) and upon discharge or death of the patient.  This set of questionnaires should be filled by medical staff in all countries for all patients that have been admitted for suspicion of COVID-19 infection.

It is of undeniable value to have a unified standardized form for data collection and subsequent analysis. However, the WHO form has two major shortcomings, it is in English and it is not in a machine readable format (unstructured PDF) so:

  1. It must be filled up manually, scanned and sent by fax or email
  2. It cannot be processed by a computer (Germany already decided to stop processing paper submissions after 7.000 patients).
  3. It cannot handle local languages nor analyze the data input in local languages in a coherent manner (Only English). If it is created in local languages in a electronic data capture system that does not provide semantic interoperability, the data cannot be aggregated across geographical areas and language barriers.
  4. The three points above mean that the data collected cannot be centrally analyzed in a time frame that will enable the data collected therein to be used to save lives (lack of semantic interoperability).

What it does

The framework for rapid data collection, analysis, visualization, and simulation in clinical settings is a set of tools:

  • We have created tool/platform to rapidly build forms that can be populated online or using data entry PDFs (still digital). We have build the WHO defined forms and have posted them online (https://covid19.nianalytics.com/crf/create?id=14573&language=en)
  • We have provided translations and local customization of the forms
  • All data points are medically coded (UMLS) and as such, all language specific forms are are feeding one data set which can easily be exported/analyzed
  • We have created a statistical model for the forms and used the built-in simulator to generate the data
  • All data can be exported via API (we support FHIR out of the box)
  • We provided a proof of concept tool for automatic visualization of data submitted via these forms http://c19.nianalytics.com/

The framework can be integrated with existing hospital information systems.

How I built it

  • Data collection through highly customizable forms (localization, flexibility, responsiveness, etc.)
  • Cloud data storage in MongoDB
  • Built-in statistical modeling and simulator
  • Visualization on web dashboard using Infogram

Challenges I ran into

A limited available development time for the hackathon :) Creating visualizations that will identify data patterns give a actionable insights to stakeholders..

What's next for InfoMaster

Development of actionable analytics that identify trends as well as correlations between collected data. Integration of additional data sources. Project completion and its use in real world settings.

Try It out

Hackathons

Technologies

.net, fhir, hl7, infogram, mongodb, restfull, snomed-ct, unified-medical-language-system

Devpost Software Identifier

254096