Deep learning-based algorithm versus physician judgement for diagnosis of myopathy and neuropathy from needle electromyography

Electromyography is a valuable diagnostic procedure for diagnosing patients with neuromuscular diseases; however, it has some drawbacks. First, diagnosis using electromyography is subjective, and in some cases, there is the potential for inter-individual discrepancies. Second, it is a time- and effort-intensive process that requires expertise to yield accurate results. Recently, a deep learning algorithm shows effectiveness for the analysis of waveform data such as electrocardiography. To overcome limitations of electromyography, we developed a deep learning-based electromyography classification system and compared the performance of our deep learning model with that of six physicians. This study included 58 subjects who underwent electromyography and were finally confirmed as having myopathy or neuropathy, or to be in a normal state between June 2015 and July 2020 at Seoul National University Hospital. We developed a one-dimensional convolutional neural network algorithm and divide-and-vote system for diagnosing subjects. Diagnosis results with our deep learning model were compared with those of six physicians with experience in performing and interpreting electromyography. The accuracy, sensitivity, specificity, and positive predictive value of the deep learning model for diagnosis as to whether subjects have myopathy or neuropathy or normal were 0.875, 0.820, 0.904, and 0.820, respectively, whereas those for the physicians were 0.694, 0.537, 0.773, and 0.524, respectively. The area under the receiver operating characteristic curves of the deep learning model for predicting myopathy, neuropathy, and normal states was better than the averaged results of six physicians. Our study showed that deep learning could play a key role in reading electromyography and diagnosing patients with neuromuscular diseases. In the future, large prospective cohort studies incorporating diverse neuromuscular diseases can enable deep learning-based electrodiagnosis on behalf of physicians. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: This study was approved by the Institutional Review Board of Seoul National University Hospital (No. 2008-055-1147). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines and uploaded the relevant EQUATOR Network research reporting checklist(s) and other pertinent material as supplementary files, if applicable. Yes All relevant data are within the manuscript and its Supporting Information files.