The Talk That Had DIA Buzzing About AI and EMR Data
Ever try searching electronic medical record (EMR) data for clinical trial inclusion and exclusion (I/E) criteria to assess protocol feasibility or identify patients to recruit? It’s difficult. 80% of the electronic medical record (EMR) consists of “unstructured data,” meaning free text of doctor’s notes and comments on medical records.
Unlike a structured diagnosis code or numerical lab result, the information needed to confirm a patient for a clinical trial might be typed in shorthand or acronyms, or need to be understood in combination with other clinical data. This is where AI comes in.
One of the trending talks from DIA’s Innovation Theater was about the value of AI applied to EMR data presented by Deep 6 AI’s VP and General Manager, Life Sciences, Jason Attanucci and Senior Solutions Consultant, Life Sciences, Allison McKinley. Here are some takeaways:
- Today, clinical trials require patients to match an average of 30 criteria for eligibility, meaning clinical research teams need to sort through hundreds or thousands of patient records to find people who qualify to participate in a trial.
- Leveraging unstructured data means that you’re not limited to ICD-10 codes and pick lists, which only account for about 20% of EMR data. With AI and natural language processing you can understand unstructured data from pathology notes, genomics reports and doctors’ notes. This allows you precision match patients to I/E criteria to assess feasibility during the study design and identify the right patients to accelerate recruitment.
- Using AI, you can assess protocol feasibility to see how modifying inclusion or exclusion criteria effects the population of patients that are available with a specific background. Or in a specific geography. Or of a specific race or ethnicity. You can see it all in real time using AI technology connected to EMR systems.
- Right now, the Deep 6 AI platform includes EMR data from major healthcare organizations across the United States, which represents over 2,000 facilities and 8,000 clinicians. This real-time data is all available to precision-match patients to specific clinical trials in any therapeutic area.
- Finally, the more real-time the data, the more precisely you can match patients. It also means you can improve pharmacovigilance by looking at adverse events that are specific to the population within the study. Deep 6 AI refreshes its EMR data consistently and it’s only getting faster with FHIR standards.
Did you miss the DIA talk? It’s not too late to reach out to Jason or Allison to learn more.