Artificial intelligence to classify ear disease from otoscopy: A systematic review and meta-analysis

Objectives

To summarise the accuracy of artificial intelligence (AI) computer vision algorithms to classify ear disease from otoscopy.

Design

Systematic review and meta‐analysis.

Methods

Using the PRISMA guidelines, nine online databases were searched for articles that used AI computer vision algorithms developed from various methods (convolutional neural networks, artificial neural networks, support vector machines, decision trees and k‐nearest neighbours) to classify otoscopic images. Diagnostic classes of interest: normal tympanic membrane, acute otitis media (AOM), otitis media with effusion (OME), chronic otitis media (COM) with or without perforation, cholesteatoma and canal obstruction.

Main outcome measures

Accuracy to correctly classify otoscopic images compared to otolaryngologists (ground truth). The Quality Assessment of Diagnostic Accuracy Studies Version 2 tool was used to assess the quality of methodology and risk of bias.

Results

Thirty‐nine articles were included. Algorithms achieved 90.7% (95%CI: 90.1–91.3%) accuracy to difference between normal or abnormal otoscopy images in 14 studies. The most common multiclassification algorithm (3 or more diagnostic classes) achieved 97.6% (95%CI: 97.3–97.9%) accuracy to differentiate between normal, AOM and OME in three studies. AI algorithms outperformed human assessors to classify otoscopy images achieving 93.4% (95%CI: 90.5–96.4%) versus 73.2% (95%CI: 67.9–78.5%) accuracy in three studies. Convolutional neural networks achieved the highest accuracy compared to other classification methods.

Conclusion

AI can classify ear disease from otoscopy. A concerted effort is required to establish a comprehensive and reliable otoscopy database for algorithm training. An AI‐supported otoscopy system may assist health care workers, trainees and primary care practitioners with less otology experience identify ear disease.

A Systematic Review of the Clinical Efficacy and Safety of CFTR Modulators in Cystic Fibrosis

Several placebo-controlled trials have been recently published evaluating novel therapies targeting the defective CFTR protein. This systematic review examines the clinical efficacy and safety of CFTR modulators in individuals with cystic fibrosis (CF) with specific genetic mutations. Online sources were searched for placebo-controlled, parallel-design clinical trials investigating CFTR modulators from January 1, 2005 to March 31, 2018. The primary outcome of interest was FEV₁% predicted (ppFEV₁). Fourteen RCTs met our eligibility criteria. The largest improvement in ppFEV₁ favouring treatment was observed for ivacaftor (IVA) in G551D individuals (≥6 years old). Both tezacaftor-ivacaftor (TEZ-IVA) and lumacaftor-ivacaftor (LUM-IVA) also improved ppFEV₁ in F508del homozygous individuals but there was increased reporting of respiratory adverse events with LUM-IVA compared to placebo. IVA also significantly improved ppFEV₁ in a sub-group of individuals ≥18 years old with an R117H mutation. No significant improvements in ppFEV₁ were observed for IVA, LUM, or TEZ in F508del homozygous individuals, LUM or LUM-IVA in F508del heterozygous individuals, or ataluren in individuals with a nonsense mutation. Significant improvements in ppFEV₁ and other clinical outcomes were observed for IVA in G551D individuals, TEV-IVA and LUM-IVA in F508del homozygous individuals, and IVA in adults with a R117H mutation.

Tissue-Engineered External Anal Sphincter Using Autologous Myogenic Satellite Cells and Extracellular Matrix: Functional and Histological Studies

The aim of the present study was to demonstrate the regaining histological characteristics of bioengineered external anal sphincters (EAS) in rabbit fecal incontinence model. The EAS of 16 rabbits were resected and decellularized. The decellularized scaffolds were transplanted to the terminal rectum following a period of 6 months of fecal incontinency (5 days after sterilization). The rabbits were divided into two groups: in group 1 (n = 8), myogenic satellite cells were injected into the transplanted sphincters. In group 2 (n = 8), the transplanted scaffolds remained in situ without cellular injection. The histological evaluation was performed with desmin, myosin, smooth muscle actin, CD31, and CD34 at 3-month intervals. The rabbits were followed for 2 years. Electromyography (EMG) with needle and electrical stimulation, pudendal and muscle electrical stimulation were also performed after 2 years of transplantation. At the time of biopsy, no evidence of inflammation or rejection was observed and the transplanted EAS appeared histologically and anatomically normal. The immunohistochemistry staining validated that the histological features of EAS was more satisfactory in group 1 in short-term follow-up. However, no statistically significant difference was detected between two groups in long-term follow-ups (p value > 0.05). In both groups, grafted EAS contracted in response to electrical signals delivered to the muscle and the pudendal nerve. However, more signals were detected in group 1 in EMG evaluation. In conclusion, bioengineered EAS with myogenic satellite cells can gain more satisfactory histological outcomes in short-term follow-ups with better muscle electrical stimulation outcomes.

The impacts of traffic-related and woodsmoke particulate matter on measures of cardiovascular health: a HEPA filter intervention study

BACKGROUND

Combustion-generated fine particulate matter (PM2.5) is associated with cardiovascular morbidity. Both traffic-related air pollution and residential wood combustion may be important, but few studies have compared their impacts.

OBJECTIVES

To assess and compare effects of traffic-related and woodsmoke PM2.5 on endothelial function and systemic inflammation (C reactive protein, interleukin-6 and band cells) among healthy adults in Vancouver, British Columbia, Canada, using high efficiency particulate air (HEPA) filtration to introduce indoor PM2.5 exposure gradients.

METHODS

We recruited 83 healthy adults from 44 homes in traffic-impacted or woodsmoke-impacted areas to participate in this randomised, single-blind cross-over intervention study. PM2.5 concentrations were measured during two consecutive 7-day periods, one with filtration and the other with 'placebo filtration'. Endothelial function and biomarkers of systematic inflammation were measured at the end of each 7-day period.

RESULTS

HEPA filtration was associated with a 40% decrease in indoor PM2.5 concentrations. There was no relationship between PM2.5 exposure and endothelial function. There was evidence of an association between indoor PM2.5 and C reactive protein among those in traffic-impacted locations (42.1% increase in C reactive protein per IQR increase in indoor PM2.5, 95% CI 1.2% to 99.5%), but not among those in woodsmoke-impacted locations. There were no associations with interleukin-6 or band cells.

CONCLUSIONS

Evidence of an association between C reactive protein and indoor PM2.5 among healthy adults in traffic-impacted areas is consistent with the hypothesis that traffic-related particles, even at relatively low concentrations, play an important role in the cardiovascular effects of the urban PM mixture.

TRIAL REGISTRATION NUMBER

http://www.clinicaltrials.gov (NCT01570062).

Animated biofeedback: an ideal treatment for children with dysfunctional elimination syndrome

PURPOSE

Animated biofeedback is an established treatment for pediatric dysfunctional voiding. Bowel dysfunction is closely associated with dysfunctional voiding. We evaluated the efficacy of animated biofeedback urotherapy in bowel and voiding dysfunction in children with dysfunctional elimination syndrome.

MATERIALS AND METHODS

A total of 80 children with dysfunctional elimination syndrome were randomly assigned to undergo animated biofeedback (group A, 40 patients) or conservative therapy (group B, 40 patients). Group A underwent animated biofeedback along with pelvic floor muscle exercises and behavioral modification (hydration, high fiber diet, scheduled voiding). Group B underwent behavioral modification only. Dysfunctional voiding symptom score, constipation and fecal soiling episodes per week (according to Paris Consensus on Childhood Constipation Terminology criteria), and uroflowmetry parameters were evaluated before and 6 and 12 months after treatment in both groups.

RESULTS

Subjective and objective voiding problems were significantly improved. Vesicoureteral reflux resolved in 7 of 9 children (78%) and urinary tract infection did not recur in 10 of 14 children (71%) within 1 year. Bladder capacity and voided volume did not significantly improve. Post-void residual and voiding time decreased considerably, while maximum and average urine flow increased significantly. All children with fecal soiling and 17 of 25 with constipation (68%) in group A were symptom-free within 1 year after treatment. Animated biofeedback therapy was more efficient than nonbiofeedback management regarding objective and subjective voiding problems and bowel dysfunction (p <0.05).

CONCLUSIONS

Animated biofeedback effectively treats bowel and voiding dysfunction in children with dysfunctional voiding. Pelvic floor muscle exercises coordinate breathing and pelvic floor muscle contractions, and are beneficial in improving bowel dysfunction.