AdvaMed

Publish Date: November 3, 2022

Read the Full Article>>

Challenges of AI in MedTech

In the medical device space, powerful insights from real-world data can seem ‘just out of reach’. Capturing patient characteristics and outcomes, at scale and with consistency over time, is challenging outside of clinical trials. Personalization – bridging the gap between large-dataset insights and what they mean for individual patients – is hardest of all. At the same time, this is where the greatest value resides.

Solutions for getting data and using it well

Existing AI and RWE tools can answer many of the questions medical device companies are asking, like how to:

  • Get the right real-world data by leveraging purpose-built real-world data networks that continuously add longitudinality and richness to cohorts.
  • Extract crucial information from data by merging incomplete, imperfect datasets to knit together fuller trajectories, and use smart technology, such as medical language processing tools, to extract key information from hard-to-access places.
  • Bridge the personalization gap by using AI tools designed for real-world data-based personalization to understand outcome patterns and connect them to individuals.

What’s Really Going on with AI?

At OM1, we’re using AI to identify distinct sets of features in real-world datasets and to associate them with outcomes. Our phenotyping tools combine strong analytic performance with medical explainability and real-world clinical utility to connect large-scale insights to the individual patient. These tools reveal patterns associated with outcomes we care about, and we can then look for these patterns in other patient groups.

In some applications, for example, a small segment of a patient population experiences a highly disproportionate share of events. Patterns distinguishing these patients can better understand who needs help when For example, we’ve found that just 4% of the type II diabetes population accounts for nearly 1 in 5 spontaneous hypoglycemic episodes requiring hospitalizations. Using our AI personalization tools, we can identify patterns associated with these patients, find them sooner, and support reducing negative outcomes through earlier and more precise interventions.

A Powerful Combination

With existing and new digital networks, medtech leaders can access high-quality data at significantly lower cost than traditional data acquisition and extract meaningful information from ‘hard to reach’ places in real-world data at scale. AI helps fill in gaps and provides insights for evidence-based decision making. Together, AI & RWD are powerful tools to support next-generation personalization and to help make diagnosis and treatment more effective at the individual level.