Insights & research

Real results, open code & research

Case Studies Open Source Research Papers

See how we take healthcare AI from proof of concept to production: deployments with measurable outcomes, open-source tools you can use today, and the research behind our methods.

Code, data and papers becoming connected insights Code · Data · Papers Open source on GitHub Validated results

Repositories

Public

dataclever /physio-signal-toolkit

Preprocessing and feature extraction for ECG, PPG and PCG signals, ready for machine learning.

pip install physio-signal-toolkit
Python ⚖ MIT
signal-processing ecg pcg
View repo
Public

dataclever /heart-sound-classifier

Reference pipeline to classify heart sound recordings as normal or abnormal.

git clone https://github.com/dataclever/heart-sound-classifier
Python ⚖ MIT
physionet deep-learning cardiology
View repo
Public

dataclever /clinical-model-eval

Metrics and reports for clinical models: sensitivity, specificity, calibration and subgroup analysis.

pip install clinical-model-eval
Python ⚖ Apache-2.0
evaluation fairness mlops
View repo

Get in touch with us

Tell us about your project. One of our engineers will reply within one business day.

Papers & Publications

Preprint arXiv:1707.04958
2017

An Ensemble Boosting Model for Predicting Transfer to the Pediatric Intensive Care Unit

J. Rubin, C. Potes, M. Xu-Wilson, J. Dong, A. Rahman, H. Nguyen, D. Moromisato
Key finding
An ensemble of AdaBoost and XGBoost on vital signs and age outperformed the pediatric early warning score on a held-out hospital.
0.85AUROC
0.73PEWS baseline
Related case · Electronic Medical Records
Conference Computing in Cardiology · PhysioNet/CinC Challenge
2016

Ensemble of Feature-based and Deep learning-based Classifiers for Detection of Abnormal Heart Sounds

C. Potes, S. Parvaneh, A. Rahman, B. Conroy
Key finding
Combines a feature-based classifier with convolutional networks on heart sound recordings; top overall score in the challenge.
0.94Sensitivity
0.78Specificity
0.86Overall
Related case · Heart Sounds
Conference IEEE EMBC 2013
2013

cortiQ – Clinical Software for Electrocorticographic Real-Time Functional Mapping of the Eloquent Cortex

R. Prueckl, C. Kapeller, C. Potes, M. Korostenskaja, G. Schalk, K.H. Lee, C. Guger
Key finding
Presents clinical software that maps eloquent cortex in real time from electrocorticography (ECoG) recordings.
Related case · Brain Electrical Activity
Journal NeuroImage · 61(4)
2012

Dynamics of Electrocorticographic (ECoG) Activity in Human Temporal and Frontal Cortical Areas During Music Listening

C. Potes, A. Gunduz, P. Brunner, G. Schalk
Key finding
In eight subjects, high-gamma ECoG activity in auditory and precentral areas tracked the loudness of music.
Related case · Brain Electrical Activity