Leveraging Artificial Intelligence To Build A Robotic Mitral Valve Program In A Community Setting
Mark Lembach, Brian Lyle, Janet Kimmel, Bryan Mahan, Daniel O'Hair.
Boulder Community Health, Boulder, CO, USA.
BACKGROUND: We sought to deliver high value mitral valve surgery in a community setting by leveraging artificial intelligence and robotics. We report a methodology for patient identification and patient outcomes following treatment.
METHODS: An artificial intelligence (AI) platform was integrated with the electronic medical record (EMR) to identify patients with severe mitral regurgitation (MR). Natural language processing and rules engines were applied to analyze structured and unstructured data from echocardiograms and the EMR for both inpatients and outpatients. Patients with severe mitral regurgitation based on quantitative and qualitative data were identified for mitral valve intervention by the platform which augmented physician referral. A focused care pathway was applied and quality metrics were recorded and analyzed.
RESULTS: Between August 1, 2019 and November 15, 2021, 18,025 echocardiograms were evaluated by the AI platform. Moderate mitral regurgitation was revealed in 2914 (16%) studies and severe MR in 546 (3%) studies. Among patients with moderate or severe MR, high risk findings including atrial fibrillation, left atrial enlargement or diminished left ventricular function were present on 515 (15%) of the studies. Heart team review identified 59 patients (30M/29F, median age 68) who underwent robotic assisted mitral valve surgery over a 21-month period. Mitral valve repair was performed in 52 patients and replacement in 7 patients. Median cross-clamp time was 100 minutes. No cases were converted to sternotomy. Postoperative atrial arrhythmia occurred in 17.2% of the cases, transfusion in 22.4%, return to the operating room 10.3%, and 5.2% required a pacemaker. There was no prolonged intubation, acute kidney injury, stroke, or post-operative mortality. Discharge to home by day 2 occurred in 44% and two patients were discharged on postoperative day 1. Thirty-day readmission rate was 8.5%. A 40% reduction in hospital stay for mitral repair and 50% reduction for replacement was achieved when compared to median STS 2021 quality metrics.
CONCLUSIONS: Artificial intelligence augments identification of patients at risk due to severe MR enabling timely repair and development of a successful robotic mitral program in a community setting.
Back to 2022 Abstracts