Houston-based Texas Children’s Hospital has developed an AI model to assess bone age in pediatric patients, allowing radiologists to allocate time to more complex tasks.
Launched in November, the AI technology interprets X-rays to estimate a patient’s bone age, which is then verified by a radiologist. This process has resulted in a 30-50% reduction in the time radiologists spend on image reading.
“By saving time on routine exams like bone age, radiologists can focus on more advanced procedures such as interventional radiology,” Anoop Vijayan, director of AI and data science at Texas Children’s, told Becker’s. “Radiology was one of the first departments to embrace automation and AI to improve efficiency and reduce workload.”
The bone age model is part of a broader effort at Texas Children’s, which has developed a dozen in-house AI solutions. These include models for employee recognition, patient no-shows, and readmissions.
For the bone age tool, Texas Children’s adapted a prebuilt AI model, validating it with their own radiology images, then built a custom solution to deploy the predictions into radiologists’ workflows.
“The AI provides the calculation, and the radiologist reviews and signs off on it,” Mr. Vijayan noted. “If there is any disagreement, we collect that feedback to further enhance our model. It’s important to emphasize that the AI does not make final decisions; a human always validates the results.”
To ensure the reliability and ethical use of AI, Texas Children’s employs a comprehensive AI governance framework. This framework is designed to validate models, evaluate performance, prevent bias, protect data privacy, and maintain transparency. The AI governance council at Texas Children’s includes leaders from clinical, operational, information security, and legal departments.
Additionally, the AI team includes radiology champions who serve as liaisons to the broader radiology group, facilitating training and monitoring the ethical use and performance of the AI model.
“The goal is to guide AI development responsibly,” said Ashok Kurian, assistant vice president of data and AI at Texas Children’s. “Involving decision-makers in AI governance is crucial to building effective and ethical AI solutions.”
