As described above, each member of dyadic diagnostic teams brings their unique perspective and skill set to address a diagnostic problem. Each of these dyads has clear benefits and limitations. The confluence of the dyads to form the patient-clinician-AI (PCA) triad may address some of the gaps left by any of the dyads alone (Figure 2).
Figure 2. Patient-clinician-AI diagnostic team
With the proliferation of AI algorithms that may inform diagnostic decision making, the structure and function of diagnostic teams are changing. For example, clinicians are receiving support from models integrated into the EHR, such as the Targeted Real-time Early Warning System (TREWS) for sepsis. A recent study of clinicians engaging with TREWS showed that clinicians perceived the system as playing a supportive role both in and beyond diagnosis.28
Clinicians will not only use AI-based diagnostic decision support embedded in EHRs, but, similar to the scenario in the Introduction, will also interact with patients who have healthcare applications that use AI algorithms. Therefore, clinicians must begin to consider how AI will inform their diagnostic decision making and how this new partner will impact their relationships with patients.