Machine learning prediction of emesis and gastrointestinal state in ferrets

Abstract

Although electrogastrography (EGG) could be a critical tool in the diagnosis of patients with gastrointestinal (GI) disease, it remains under-utilized. The lack of spatial and temporal resolution using current EGG methods presents a significant roadblock to more widespread usage. Human and preclinical studies have shown that GI myoelectric electrodes can record signals containing significantly more information than can be derived from abdominal surface electrodes. The current study sought to assess the efficacy of multi-electrode arrays, surgically implanted on the serosal surface of the GI tract, from gastric fundus-to-duodenum, in recording myoelectric signals. It also examines the potential for machine learning algorithms to predict functional states, such as retching and emesis, from GI signal features. Studies were performed using ferrets, a gold standard model for emesis testing. Our results include simultaneous recordings from up to six GI recording sites in both anesthetized and chronically implanted free-moving ferrets. Testing conditions to produce different gastric states included gastric distension, intragastric infusion of emetine (a prototypical emetic agent), and feeding. Despite the observed variability in GI signals, machine learning algorithms, including k-nearest neighbors and support vector machines, were able to detect the state of the stomach with high overall accuracy (>75%). The present study is the first demonstration of machine learning algorithms to detect the physiological state of the stomach and onset of retching, which could provide a methodology to diagnose GI diseases and symptoms such as nausea and vomiting.

Selective stimulation of the ferret abdominal vagus nerve with multi-contact nerve cuff electrodes

“…In the additional three ferrets, an identical procedure was used to implant a four-contact nerve-cuff electrode around the ventral abdominal vagus nerve and four planar electrodes, each with four contacts (Micro Leads, Inc. Somerville, MA), were sutured to the ventral gastric surface. The locations of these planar electrodes on the ventral gastric surface was similar to our prior study 37 .…”

“…Moreover, it is not well documented but isoflurane is reported to affect gastric myoelectric activity 36 . There are also substantial physiological differences between animals, which we have observed in in gastric myoelectric activity in awake and anesthetized ferrets 37 . In this study, we quantified selectivity via CAP recordings in one set of ferrets and measured changes in GI myoelectric activity with similar stimulation parameters in a different set of animals.…”

“…We used recordings of the nodose ganglia to assess selectivity but it is unlikely this approach could be applied clinically; therefore, it will be important to use more accessible targets, such as serosal gastric myoelectric activity or cuff electrode recordings from the vagus nerve to confirm VNS selectivity [e.g. 37 , 38 ]. Such an approach may lead to the application of closed-loop techniques to record and apply VNS to treat diseases of the abdominal cavity, such as GI disease, obesity, diabetes, and inflammatory disorders.…”

“…We have previously shown that GI myoelectric activity can be used to identify the physiological state (normal, distended, pre-retch) of the stomach 37 . For this study, we analyzed data from a subset of those experiments to demonstrate the selective effect of VNS on GI myoelectric activity.…”

“…For every planar electrode, the waveform recorded on each of the four contacts was averaged to generate a single GI myoelectric waveform for that electrode. Analysis methods for GI myoelectric activity were adopted from our prior study 37 . Briefly, each planar-averaged GI myoelectric signal was filtered using a low-pass Butterworth filter with a 2.5 Hz (150 cpm, 4th order) cutoff.…”