Purpose: For patients undergoing gastrointestinal surgery, postoperative ileus (POI) is difficult to predict and occurs at rates up to 30%. We hypothesized that discrete gastrointestinal acoustic biomarkers correlating to POI development may present in early postoperative hours. To test this hypothesis, we conducted a two-phase investigation to record and analyze intestinal sounds, then prospectively test the feasibility of predicting POI using a noninvasive device trained with an algorithm developed from Phase 1 results.
Methods: In Phase 1, a library of intestinal sounds from normal volunteer subjects and patients undergoing intestinal resection were recorded. An acoustic biomarker most correlated with POI was identified and used to develop a predictive algorithm. In Phase 2, an autonomous, wearable device possessing listening and signal processing capability was developed and prospectively tested for prediction of POI in patients undergoing bowel resection.
Results: In Phase 1, 30 patients undergoing resection were studied, 9 of whom developed POI, from which an acoustic biomarker was identified and analyzed. In Phase 2, prospective prediction of POI in 75 enrolled subjects undergoing bowel resection was assessed. POI was correctly predicted in 12 of 14 subjects who developed POI; prediction that POI would not occur was correct in 39 of 51 subjects. Sensitivity and specificity were 85.7% and 63.9%, respectively. Negative predictive value and accuracy were 95.1%, and 68%, respectively.
Conclusions: Our study demonstrates feasibility of POI prediction based on gastrointestinal sounds using a noninvasive device. This device may help risk stratify patients likely to develop POI.