AI does just as well as doctors when it comes to predicting postoperative complications
Artificial intelligence is invading all areas of our lives, and now it has a toehold in predicting postoperative complications, showing the same level of success as doctors.
A study reported in JAMA Network Open looked at more than 74,000 inpatient surgical procedures involving more than 58,000 patients to see how well artificial intelligence could predict postoperative complications when given EHR and access to a surgeon’s mobile device output.
The AI platform ended up performing at the same level as a surgeon. Researchers say the study illustrates that if AI is given the right data and integrated into the clinical workflow, it has the potential to augment surgical decision-making. The AI can help surgeons decide on the appropriateness of procedures, targeting risk-reduction strategies, and helping with postoperative resources.
The application output displays patient information, risk of postoperative complications, top 3 risk factors for each complication, and patterns of complications for individual surgeons compared with their colleagues.
The potential cost-savings is substantial, according to the research. There are more than 15 million inpatient surgical procedures performed annually, with complications occurring in as many as 32%, increasing costs by up to $11,000 per major complication. Cognitive and judgment errors are major sources of potentially preventable complications. For example, underestimation of the risk of complications may be associated with postoperative undertriage of high-risk patients to general wards rather than intensive care units and an increased prevalence of hospital deaths.
Researchers say that a high-performance, data-based clinical decision support has the potential to mitigate harm from cognitive errors occurring when estimating the risk of postoperative complications. All patients have a unique risk profile that is specific to their demographic characteristics, comorbid conditions, physiological reserve, planned surgical procedure, and surgeon’s skill; clinicians have had mediocre performance in estimating risk probabilities. The more data the AI is given, the better its performance.
The current crop of decision support tools are intended to augment these estimations, but many are hindered by time-consuming manual data entry requirements and lack of integration with clinical workflow.