Identifying Disease of Interest With Deep Learning Using Diagnosis Code

Background@#Autoencoder (AE) is one of the deep learning techniques that uses an artificial neural network to reconstruct its input data in the output layer. We constructed a novel supervised AE model and tested its performance in the prediction of a co-existence of the disease of interest only using diagnostic codes. @*Methods@#Diagnostic codes of one million randomly sampled patients listed in the Korean National Health Information Database in 2019 were used to train, validate, and test the prediction model. The first used AE solely for a feature engineering tool for an input of a classifier. Supervised Multi-Layer Perceptron (sMLP) was added to train a classifier to predict a binary level with latent representation as an input (AE + sMLP). The second model simultaneously updated the parameters in the AE and the connected MLP classifier during the learning process (End-to-End Supervised AE [EEsAE]). We tested the performances of these two models against baseline models, eXtreme Gradient Boosting (XGB) and naïve Bayes, in the prediction of co-existing gastric cancer diagnosis. @*Results@#The proposed EEsAE model yielded the highest F1-score and highest area under the curve (0.86). The EEsAE and AE + sMLP gave the highest recalls. XGB yielded the highest precision. Ablation study revealed that iron deficiency anemia, gastroesophageal reflux disease, essential hypertension, gastric ulcers, benign prostate hyperplasia, and shoulder lesion were the top 6 most influential diagnoses on performance. @*Conclusion@#A novel EEsAE model showed promising performance in the prediction of a disease of interest.

AI-powered medical devices for practical clinicians including the diagnosis of colorectal polyps

The integration of medical devices with artificial intelligence (AI) software is rapidly advancing as technology progresses. AI machine learning can be used in commercial medical services to generate practical data; there is evidence that it can be integrated into newly developed devices. However, such devices must undergo approval, regulation, and supervision. The Food and Drug Administration approves regulations for numerous machine-learning medical devices and shares open lists with the public. In this article, we examine recent medical AI devices in different fields, including the diagnosis of colorectal polyps.Current Concepts: Currently, in the field of gastroenterology, there has been a significant amount of research aimed at enhancing adenoma detection rates using tools powered by AI, such as the EndoScreener and GI Genius. Various such devices have also been developed for other fields; examples include the 23andMe Personal Genome Service for DNA detection, Spectral MD’s DeepView platform for wound imaging in surgery, Gili Pro BioSensor for monitoring vital signs, DreaMed Advisor Pro for diabetes, Minuteful for urinary analysis, BrainScope TBI for cerebral diagnosis, Compumedics Sleep Monitoring System for sleep disorders, Idx-DR v2.3 for ophthalmology, and EarliPoint system for pediatrics.Discussion and Conclusion: By the time this article is published, it is likely that even more AI medical devices will have been approved and commercialized. The development of such devices should be strongly encouraged. Additionally, we anticipate greater involvement from practitioners in the development and validation of diverse medical AI devices in Korea.

The number of resident physicians and the mortality rate in a medical intensive care unit / 대한내과학회지

BACKGROUND/AIMS: The treatment outcome of patients hospitalized in intensive care units (ICUs) can be influenced by physician factors, including both intensivists and resident physicians. We evaluated the association between the number of residents who are exclusively responsible for the ICU and the mortality rate in a medical ICU. METHODS: The data obtained from an open medical ICU in a teaching hospital from Jan. 2005 to Dec. 2009 were analyzed retrospectively. We evaluated the associations between the ICU mortality rate and both the number of resident physicians and the number of patient-days per resident physician using multivariate Poisson regression analysis adjusted for year and month. RESULTS: The months with fewer than two residents tended to have a higher ICU mortality rate, although this difference was not significant in the univariate analyses. Multivariate Poisson regression analysis showed that months with fewer than two residents had a significantly higher ICU mortality rate compared with months with two residents (incidence risk ratio (IRR) 1.59, 95% confidence interval (CI) 1.05-2.41; p=0.029). The number of ICU patient-days per resident physician was not associated with the ICU mortality rate (IRR; 1.00, 95% CI, 0.99-1.01; p=0.649). CONCLUSIONS: The presence of fewer than two residents exclusively responsible for the medical ICU was an independent risk factor of a higher ICU mortality rate. However, no association was found between the number of ICU patient-days per resident physician and the ICU mortality rate.

Updates on Enterovirus Surveillance in Korea / 감염과화학요법

PURPOSE: We identified the causative viruses from patients with aseptic meningitis, acute hemorrhagic conjunctivitis and other enterovirus-related diseases to understand the epidemiological patterns and prevailing strains of enterovirus infections each year. MATERIALS AND METHODS: During 1999-2003, we examined 3,260 specimens from 2,939 patients with aseptic meningitis or other clinical manifestations for the presence of enteroviruses by using both cell culture/ neutralisation test and reverse transcription-polymerse chain reaction-sequencing. To investigate the etiological agents which caused an epidemic of acute haemorrhagic conjunctivitis, conjunctival swab samples from acute haemorrhagic conjunctivitis patients showing cytopathic effects in HEp2 cells were tested by enteroviral specific PCR. RESULTS: We identified 603 isolates of enteroviruses (20.5%) among 2,939 cases and 22 serotypes of human enteroviruses were isolated during this 5 year period. Echovirus 13 and coxsackievirus A24 in 2002 and coxsackievirus A9 in 2003 were the first enterovirus to be indentified in Korea since we began the enterovirus surveillance in 1993. While an epidemic of echovirus 13 infection in Korea began in Gwangju and Jeolla province in 2002 and spread to Seoul, Gyunggi, Busan, Ulsan and other regions, echovirus 6 isolates in 2002 were mainly detected in Busan specimens and some Gwangju samples. From the nucleotide sequencing of enteroviral PCR products of conjunctival swab specimens, we found 85% nucleotide homology to coxsackievirus A24 (D90457). CONCLUSIONS: We isolated 603 enteroviral isolates among 2939 cases during 1999-2003. Echovirus 13 and coxsackievirus A24 were the first enterovirus to be identified in Korea and caused nationwide epidemics in 2002.

Updates on Enterovirus Surveillance in Korea / 감염과화학요법

PURPOSE: We identified the causative viruses from patients with aseptic meningitis, acute hemorrhagic conjunctivitis and other enterovirus-related diseases to understand the epidemiological patterns and prevailing strains of enterovirus infections each year. MATERIALS AND METHODS: During 1999-2003, we examined 3,260 specimens from 2,939 patients with aseptic meningitis or other clinical manifestations for the presence of enteroviruses by using both cell culture/ neutralisation test and reverse transcription-polymerse chain reaction-sequencing. To investigate the etiological agents which caused an epidemic of acute haemorrhagic conjunctivitis, conjunctival swab samples from acute haemorrhagic conjunctivitis patients showing cytopathic effects in HEp2 cells were tested by enteroviral specific PCR. RESULTS: We identified 603 isolates of enteroviruses (20.5%) among 2,939 cases and 22 serotypes of human enteroviruses were isolated during this 5 year period. Echovirus 13 and coxsackievirus A24 in 2002 and coxsackievirus A9 in 2003 were the first enterovirus to be indentified in Korea since we began the enterovirus surveillance in 1993. While an epidemic of echovirus 13 infection in Korea began in Gwangju and Jeolla province in 2002 and spread to Seoul, Gyunggi, Busan, Ulsan and other regions, echovirus 6 isolates in 2002 were mainly detected in Busan specimens and some Gwangju samples. From the nucleotide sequencing of enteroviral PCR products of conjunctival swab specimens, we found 85% nucleotide homology to coxsackievirus A24 (D90457). CONCLUSIONS: We isolated 603 enteroviral isolates among 2939 cases during 1999-2003. Echovirus 13 and coxsackievirus A24 were the first enterovirus to be identified in Korea and caused nationwide epidemics in 2002.