253 Better and faster automatic sleep staging with artificial intelligence

Introduction Sleep staging of polysomnography (PSG) is a time-consuming task, it requires significant training, and significant variability among scorers is expected. A new software (MEBsleep by Medibio Limited) was developed to automatically perform sleep scoring based on machine learning algorithms. This study aimed to perform an extensive investigation of its agreement with expert sleep technicians. Methods Forty polysomnography recordings of patients that were referred for sleep evaluation to three sleep clinics were retrospectively collected. Three experienced technicians independently staged the recording complying with the scoring rules of the American Academy of Sleep Medicine guidelines. Positive Percent Agreement (PPA), Positive Predictive Value (PPV), and other agreement statistics between the automatic and manual staging, among the staging performed by the three technicians, and their differences were calculated. Bootstrap resampling was used to calculate 95% confidence intervals and statistical significance of the differences. Results Automatic staging took less than two minutes per PSG on a consumer laptop. The automatic staging resulted for the most comparable (PPA difference of N1, N3, and REM; PPV difference of N1, N2, N3, and REM) or statistically significantly more in agreement with the technicians’ staging than the between-technician agreement (PPA difference of N2: 3.90%, 95% bootstrap CI 1.79%-6.01%; PPV difference of Wake: 1.16%, 95% bootstrap CI 0.64%/1.67%), with the sole exception of a partial reduction in the positive percent agreement of the Wake stage (PPA difference of Wake -7.04%, 95% bootstrap CI -10.40%/-3.85%). The automatic staging also demonstrated very high accuracy in an indirect comparison with other similar software. Conclusion Given these promising results, the use of this software may support sleep clinicians by improving efficiency in sleep scoring. Support (if any):

Imaging-Based Staging of Hepatic Fibrosis in Patients with Hepatitis B: A Dynamic Radiomics Model Based on Gd-EOB-DTPA-Enhanced MRI

Accurate grading of liver fibrosis can effectively assess the severity of liver disease and help doctors make an appropriate diagnosis. This study aimed to perform the automatic staging of hepatic fibrosis on patients with hepatitis B, who underwent gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging with dynamic radiomics analysis. The proposed dynamic radiomics model combined imaging features from multi-phase dynamic contrast-enhanced (DCE) images and time-domain information. Imaging features were extracted from the deep learning-based segmented liver volume, and time-domain features were further explored to analyze the variation in features during contrast enhancement. Model construction and evaluation were based on a 132-case data set. The proposed model achieved remarkable performance in significant fibrosis (fibrosis stage S1 vs. S2–S4; accuracy (ACC) = 0.875, area under the curve (AUC) = 0.867), advanced fibrosis (S1–S2 vs. S3–S4; ACC = 0.825, AUC = 0.874), and cirrhosis (S1–S3 vs. S4; ACC = 0.850, AUC = 0.900) classifications in the test set. It was more dominant compared with the conventional single-phase or multi-phase DCE-based radiomics models, normalized liver enhancement, and some serological indicators. Time-domain features were found to play an important role in the classification models. The dynamic radiomics model can be applied for highly accurate automatic hepatic fibrosis staging.

0442 Experimenting Automatic Sleep Analysis Application in a Clinical Context

Introduction Multiplication of publications describing groundbreaking automatic sleep analysis processes and algorithms push for real-life experimentation in clinical context, outside of controlled research environments. Methods Various automatic sleep analysis processes from the literature were implemented and orchestrated in a streamlined workflow. Artificial Intelligence algorithms using regular statistical learning or deep learning were re-trained on our own data after repeating the ad-hoc pre-processing steps described in the corresponding articles. For this, we used polysomnographic records previously taped in our clinic, subject to adequate legal authorizations and agreements: 500 nights from single patients with various pathologies. Those trained models were then applied to newly recorded polysomnographies through a platform developed and hosted on premise. For each polysomnography, a standardized and automatized report were generated and transmitted to the clinician in charge of the analysis. This report contains algorithms outputs, including automatic staging and related statistics such as hypnodensity, quantitative electroencephalography (EEG) analysis, spindles detection and automatic diagnosis. Aggregated record statistics are displayed next to our database statistics for benchmarking purposes. Results For sleep staging, we not only reproduced the results of the selected literature but obtained better metrics: a 0.76 Kappa agreement vs 0.69 in the literature. This may be due to our larger training database or the quality of physiologic signals in our data. Clinicians showed interest in the automatic staging part of the analysis. They noticed algorithm errors are mostly focused on ambiguous epochs, just like visual scoring. However, they found help into automated output and explanatory variables (hypnodensity) to score those ambiguous epochs. Conclusion Automatic sleep analysis algorithms used as decision helping tools shows real potential and should be generalized, as long as underlying processes are published and understood by users and clinicians. Support Banque Publique d’Investissement.

Automatic Staging via Partial Evaluation Techniques

Partial evaluation and staging are two of the well-known symbolicmanipulation techniques of programs which generate efficientspecialized code. On one hand, partial evaluation provides us with anautomatic means to separate a program into two (or more) stages butits behavior is perceived as hard to control. On the other hand,staging (or staged calculus) requires us to manually separate aprogram but with full control over its behavior. In the previouswork, the first author introduced a framework to relate the twotechniques, giving a unified view to the two techniques. In thispaper, we extend the framework to handle the cross-stage persistence(CSP) and show that the 2-level staging annotation obtained by theautomatic separation is the best staging annotation in a system whereCSP is allowed for base-type values only. In the presence of CSP forhigher-type values, on the other hand, there is no single annotationthat is better than all the other annotations.