Sacral Neuromodulation — Database for AI/ML-based therapy design for monitoring and outcomes analysis — 84a — Janae Hahn, Dane Reeves, Bichar Dip Shrestha Gurung, Dr. Dwight Nelson, Dr. Matthew A. Barker, Dr. Etienne Gnimpieba

Sacral neuromodulation is an FDA approved therapy delivered using an implanted medical device. Effective therapy treats urinary and bowel symptoms relating to incontinence. This well established therapy is delivered by an implanted pulse generator with electrical leads that target  sacral nerves. While significant engineering has gone into the development of the therapy, there remains too little scientific guidance for clinicians about device programming and there is inadequate quantitative data available to drive therapy improvements. Our project will use the  tools of artificial intelligence (AI) and machine learning (ML) to improve the clinical and research  use of the widely dispersed and often informal clinical and electronic data that are spread across  several platforms. 

Our USD team has met with the Avera clinical team and surgeon to observe patients and  care teams using, programming, and taking measurements from the devices. We are writing an  Avera IRB clinical protocol and designing secure processes and databases to capture essential  patient and clinical data in a HIPPA-compliant manner. Preliminary observations suggest the clinic generates >100 different variables/patient that can be tracked over >7 office, surgical, and follow up visits. The data are diverse measurements, highlights of patient questionnaires, and  transcriptions of verbal clinical descriptions. Values are stored as note fields or PDFs in >4 databases and different physical locations including on-patient, in the clinic, in Avera medical  records and sometimes corporate databases.

In the short term, improvements in data automation will help research teams access  quantitative data and allow them to test hypotheses regarding the therapy, device programming and patient outcomes. Longer term, we hope this project will lead to neuromodulation devices that  are easier to implant, simpler for clinics to use and monitor, and self-programming; Ultimately  making them more effective for patient therapy.

University of South Dakota
Dr. Etienne Gnimpieba