Artificial Intelligence has been hailed as the most recent digital health development that can help healthcare systems do more with constrained resources. But is that the case?

A study was conducted to assess the impact of an AI-enabled triage on labour productivity in Sweden. A dual-centre, retrospective study comparing labour productivity of different staff groups in Swedish primary care through a differences-in-differences analysis was conducted. Quantitative data was extracted from the electronic health record systems of the treatment and control practices. Data included the number of total, planned, and unplanned appointments for all patients aged 16 and over, full time equivalents of doctors, nurses and administrators, and total practice list sizes over the same period of 11 months. Collected data sets were analysed using STATA software. Additionally, an interview was conducted with a General Practitioner Partner of the treatment practice to obtain qualitative data on the impact of the triage system. Finally, the urgency distribution for all cases submitted through the system at the treatment practice during the study period was extracted from the data management systems. 

On average, doctors at the health center were able to conduct 230.50 additional total appointments per full time equivalent (p<0.01). This broke down to 78.12 more planned appointments and 152.40 unplanned appointments per full time equivalent doctor (p<0.05). Nurses conducted 112.80 additional total appointments per full time equivalent, of which 46.94 were additional planned appointments per full time equivalent (p<0.05). The were no significant increases per full time equivalent administrators. 

The results suggest that AI-enabled triage implementation resulted in an improvement in labour productivity for clinicians in primary care. Thus, AI could be implemented to support healthcare systems to overcome resource challenges!