Imagine if decisions about your clinical care were supported by artificial intelligence (AI). While introducing AI into high-stakes medical scenarios may seem risky, these tools have the potential to outperform human doctors in predicting certain medical outcomes. They can account for the complexity of each patient’s circumstances while also reducing costs for those who may not require specialized care.
Consider the threat of stillbirth, a sadly common outcome affecting about one in 160 pregnancies nationwide. For individuals at higher risk, intensive monitoring and medical testing offer the best chances of delivering a healthy baby. However, much remains unknown about the causes of stillbirth, making it challenging to identify which pregnancies are at risk.
Dr. Nathan Blue, an assistant professor specializing in obstetrics and gynecology at the University of Utah, notes the clinical uncertainty surrounding risk factors for stillbirth, especially concerning fetuses with lower weights. Even for specialists like Dr. Blue, determining which patients require heightened medical monitoring is difficult due to the multifaceted nature of stillbirth risk factors.
In response to this complexity, artificial intelligence emerges as a promising solution. AI can analyze vast databases of past pregnancy outcomes to identify patterns and predict outcomes more accurately and swiftly than human doctors alone. Dr. Blue is developing an AI-based tool that will sift through extensive pregnancy data to discern hidden warning signs of stillbirth, ranging from genetic and environmental factors to clinical measurements.
This AI tool could estimate the risk of stillbirth for future pregnancies by considering individual risk factors. When a patient presents with a fetus smaller than expected, their doctor could input their unique characteristics into the AI tool to calculate a personalized risk estimate of stillbirth.
Armed with this knowledge, patients and doctors can make informed decisions about next steps. Those with high-risk pregnancies can receive closer monitoring, while those at lower risk may avoid unnecessary medical procedures, thereby alleviating stress and reducing costs.
However, AI tools are not without risks. They can be influenced by biases present in real-world data, such as gender- and race-based biases affecting past patterns of care. Dr. Blue believes transparency regarding the factors influencing predictions can help identify and address biases, ultimately promoting equitable care.
While concerns about AI exist, Dr. Blue emphasizes that good clinical care extends beyond a single number or prediction. Regardless of whether risks are calculated by AI, equations, or a doctor’s judgment, each patient’s unique priorities and values must be considered.
In the realm of high-risk pregnancies, where predicting outcomes is uncertain even for experienced doctors, AI offers a potential path to improved decision-making and personalized care. As AI tools continue to evolve, they may work alongside doctors to enhance diagnoses and provide patients with comprehensive, personalized information to support their decisions.
