Exploring new subgroups for irritable bowel syndrome using a machine learning algorithm

Irritable bowel syndrome (IBS) is a complicated gut-brain axis disorder that has typically been classified into subgroups based on the major abnormal stool consistency and frequency. The presence of components other than lower gastrointestinal (GI) symptoms, such as psychological burden, has also been observed in IBS manifestations. The purpose of this research is to redefine IBS subgroups based on upper GI symptoms and psychological factors in addition to lower GI symptoms using an unsupervised machine learning algorithm. The clustering of 988 individuals who met the Rome III criteria for diagnosis of IBS was performed using a mixed-type data clustering algorithm. Nine sub-groups emerged from the proposed clustering: (I) High diarrhea, pain, and psychological burden, (II) High upper GI, moderate lower GI, and psychological burden, (III) High psychological burden and moderate overall GI, (IV) High constipation, moderate upper GI, and high psychological burden, (V) moderate constipation and low psychological burden, (VI) High diarrhea and moderate psychological burden, (VII) moderate diarrhea and low psychological burden, (VIII) Low overall GI, and psychological burden, (IX) Moderate lower GI, and low psychological burden. The proposed procedure led to the discovery of new homogeneous clusters in addition to certain well-known Rome sub-types for IBS.

Automated summary

This research uses an unsupervised machine learning algorithm to redefine subgroups of irritable bowel syndrome (IBS) based on upper gastrointestinal symptoms and psychological factors in addition to lower gastrointestinal symptoms. It discovered new homogeneous clusters in addition to well-known Rome sub-types for IBS.