Skip to main content

Table 4 Results of individual studies

From: The potential of computerised analysis of bowel sounds for diagnosis of gastrointestinal conditions: a systematic review

Paper

Target condition

Main findings

Craine et al. 1999 [13]

Irritable bowel syndrome (IBS)

Fasting s-s interval useful: significant difference between IBS and healthy individuals ((t test) p < 0.0001).Using 640 msec as the cut-off, sensitivity was 89% and specificity was 100% on the preliminary data (AUC = 0.99)

Craine et al. 2001 [14]

IBS and Crohn’s Disease

Useful: fasting s-s interval is higher in Crohn’s and healthy individuals than in IBS individuals (heterogeneity across 3 groups (Kruskal Wallis) p < 0.0001). Using an s-s interval of 740 msec gave a sensitivity (NPV) of 97.8%, and TPV of 13.5% for distinguishing between IBS and controls (AUC = 0.978). The AUC for distinguishing CD from IBS patients was 0.843. High s-s interval in an individual with IBS symptom should prompt a search for an alternative diagnosis such as Crohn’s. Unable to differentiate between healthy and Crohn’s individuals based on this feature. The AUC for distinguishing CD from controls was only 0.709.

Craine et al. 2002 [15]

IBS and non-ulcer dyspepsia (NUD)

Useful: significant differences across all groups in s-s interval (Kruskal Wallis) p < 0.0001. Control vs IBS significantly different in % power in lower freq sounds, especially in right lower quadrant (RLQ) (p < 0.001). Also, significant differences between NUD and controls in ratio of gastric sounds to RLQ sounds (p < 0.001). Fewer differences between FGID groups, but IBS and NUD patients significant differences in ratio of gastric to RLQ sounds (p < 0.001). Note, the authors split the NUD patients into two groups based on s-s interval.

Hadjileontiadis et al. 1999 [16]

IBS, diverticular disease (DD), bowel polyp (2 cm) and ulcerative colitis (UC)

Useful: limited statistics, but scatter plots of HOC using the optimum HOC domain discriminate between patients and controls. The φ2 (non-weighted) statistic or with weights adapted to the HOC with maximum discriminative information, provides another simple discriminative feature between controls and DD and between DD and UC.

Yoshino et al. 1990 [17]

Intestinal obstruction (large and small bowel)

Useful: objective indicator of surgery for intestinal obstruction. Seriousness could be identified from bowel sounds characteristics—objective measure, and suggests treatment regimen—conservative or operative. Seriousness order: sounds type3 > sound type 2 > sound type 1.Those with type three sounds all had strangulating obstructions or a condition requiring surgery. Fewer of those with type 2 sounds required surgery (after a longer delay than group3 cases) and all of those with type 1 sounds were simple obstructions which did not require surgery. Upper and range of sound frequencies were higher significantly higher in type 1 than normal (p < 001). Peak (p < 0.001) and upper (p < 0.01) frequencies were higher in type 2 relative to type 1. Peak, upper, and range was significantly higher in type 2 relative to normal (p < 0.001).Type 3 significantly different from normal in peak (p < 0.001) and upper (p < 0.001) freq. Type 3 is significantly different from type1 in peak freq (p < 0.01), but there were no significant differences between type 1 and type 2 in sound frequencies.

Ching et al. 2012 [18]

Small and large bowel obstruction

Non-specific for diagnosing bowel obstruction. No sig diffs between the 3 groups (no obstruction, subacute, acute) in sound to sound interval, sound duration, dominant freq, and peak freq when look at all cases. However, incidence of prolonged bowel sounds increased significantly across the 3 groups in the suspected large bowel cases (p = 0.025). The bowel sounds may be useful in locating the site of an acute obstruction. Sound duration (p = 0.021) and the dominant frequency (p = 0.003) were significantly higher in large bowel obstruction vs small bowel obstruction. No bowel sound feature correlated with bowel calibre. Some indication of severity: sound to sound interval longer in the small bowel obstruction group that underwent surgery (p < 0.01).

Sugrue and Redfern 1994 [19]

Acute abdomen, varying severity (appendicitis, cholecystitis and intestinal obstruction)

Useful: mean number of bowel sounds was greater in normal subjects than those with appendicitis (p < 0.05) and obstruction (p < 0.05). Bowel sounds not significantly different in length for appendicitis and controls. However, sound to silence ratio was less in appendicitis (more silence) than in controls (0.05). Sounds significantly longer in cholecystitis and intestinal obstruction than in controls and those with appendicitis (p < 0.05).

Kim et al. 2011A [20]

Delayed gastric emptying

Useful: this method could be used for the non-invasive measurement of bowel motility. Jitter and shimmer of the bowel sounds of healthy group members were higher than those with spinal cord injury. Correlation coefficient between CTTs and eCTT was 0.987 (S.E. = 7.99 h)

Kim et al. 2011B [21]

Delayed gastric emptying

Useful: bowel sound features could be clinically useful for measurement of bowel motility. Jitter and shimmers of normal subjects were significantly higher than patients (p < 0.01). Performance of the algorithm: 12 random feature datasets used to train the model and 6 datasets used to test the algorithm. Outcome: correlation coefficient between CTT and eCTT was 0.89 (mean average error = 10.6 h). Estimation errors slightly better than the regression model derived from this data (similar to that used in Kim et al. 2011A).

Tomomasa et al. 1999 [22]

Pyloric stenosis and impaired gastric emptying in infants

Useful: decreased gastrointestinal sounds are suggestive of HPS and a useful indicator of gastric emptying and bowel motility after pylormyotomy. Mean SI was significantly less in pyloric stenosis patients before surgery than in healthy controls (p = 0.0013). Incidence of post-op symptoms negatively correlated significantly with SI at 24 h post-op (p = 0.035, R2 = 0.373). There was a significant positive correlation between SI and gastric emptying (p = 0.018).

Spiegel et al. 2014 [23]

Post-operative ileus

Useful: there is a relationship between intestinal rate and post-op clinical status. Significant differences between the three groups. However, there is some overlap between POI and re-feeding group, so only indicative. ROC analysis on differentiation of healthy controls (not the re-feeding group) and the POI group revealed a threshold of 0.1 events per second to give an AUC of 0.995.

Kaneshiro et al. 2016 [24]

Post-operative ileus

Useful: for the 5 day post-op period, intestinal rate (IR) was significantly lower in the POI group. Drop in IR between POD 1 and 2 observed in the POI group was sign diff from the increase seen in the non-POI group. % time IR was below the 5th percentile, also differed significantly. Used these last two variables to predict POI. ROC area under the curve was 0.83. Using a test threshold of 0.4, able to differentiate between groups with sensitivity 63%, specificity 72% and NPV 83%. High NPV suggests use of negative test result as a rule-out-tool for POI to aid decision making around diet advancement.

Campbell et al. 1989 [25]

Diarrhoea—severe (post-gastrectomy) and mild idiopathic

Limited usefulness: SVA significantly greater in the severe diarrhoea group than the healthy controls (p < 0.01). Difference not significant between mild and severe and mild and controls. Inverse correlation between SVA energy value and OCTT p < 0.01 (Spearman’s rho = −  0.486). Drug stimulation of the GI tract caused a significant increase in SVA measurements.

Liatsos et al. 2003 [26]

Small volume ascites

Useful: novel diagnostic features of bowel sounds identified that could give rise to a new diagnostic tool in routine clinical practice. There was a distinct separation of all cirrhotic patients with small ascites from controls (p < 0.0001). Coincided with radiological findings.