HITCON SHOWCASE

FIND MORE INFORMATION HERE

https://www.sciencedirect.com/science/article/pii/S0169260722002814

https://www.frontiersin.org/articles/10.3389/fcvm.2022.893090/full

PLEASE FIND VIDEO DETAILS

https://youtu.be/Kc5He693c1s

Doctors-Engineers team up to develop an app and Device to prevent stroke

by early detection of irregular heart rhythms.

Heart rhythm abnormalities, such as Atrial Fibrillation, if not detected early , can potentially lead to embolic phenomena like stroke. Due to their asymptomatic and sporadic nature, it is difficult to detect them in routine in-hospital checkups at the early stages. Thus, it requires continuous monitoring in the

patient’s free-living conditions. Though many new wearable technologies/smartwatches allow taking

short (30-60 seconds) ECG under free-living conditions, they are inadequate if rhythm abnormalities are

asymptomatic and do not occur when those short recordings are being taken periodically. Although many

smartwatches offer continuous heart rate monitoring using a photoplethysmography (PPG) sensor, ECGremains the gold standard for arrhythmia monitoring.

Manual analysis of days or months-long continuous ECGs for detecting irregular heart rhythms is an arduous task and often limited by the lack of resources.

In recent years, many AI and non-AI-based automatic ECG analysis algorithms have been developed;

however, they produce a significant number of false positives when used on patient-operated ECG fromfree-living conditions and require manual re-evaluation. Also, ECG analysis without an understanding ofthe patient's ambulatory contexts is prone to misclassification as sometimes motion artifacts also may mimic the arrhythmias and could lead to over diagnosis.

To address these challenges, Dr Kamal Sharma , Ahmedabad based senior international cardiologist along

with Dr Devender Singh from Technical University of Denmark and His colleagues developed a context-

aware ECG collection app (mCardia) and AI-based algorithm (DeepAware) for continuous and

longitudinal analysis of heart ECG under patients' free-living conditions. Alone with the continuous ECG,they collected the patient's ambulatory contexts (such as activities, movement acceleration, bodypositions, symptoms, and food intake) and use them to improve the DeepAware-- our AI-based rhythm abnormality detection algorithm. Along with the features of RR variability and P wave morphology, they also utilized context awareness that significantly improved the false positive rates. They collected over5000 hours of continuous ECG data from Indian patients to train and evaluate their algorithm. 

DeepWare achieved an accuracy of 98.06% on our ambulatory datasets and also outperformed state-of-the-art on other public arrhythmia datasets like MIT AFDB and MIT BHI (Accuracy of 98.62% and 91.82%,respectively). Since no public arrhythmia datasets are available from the Indian population, they have alsomade their CACHET-CADB publicly available for the AI research community to benchmark and buildnew AI models.

 As data is the “fuel” for AI, it is essential that AI algorithms to be used in the Indian context are trained and evaluated on data from the Indian population. Many existing AI-based rhythm abnormality detection algorithms trained on public arrhythmia datasets, such as MIT AFDB, and MIT BHI, show good performance on these datasets (or other clean in-hospital ECG datasets). 

However, as demonstrated in their 4 published research articles, these other datasets give higher false positives when used on patient-operated ECG under free-living conditions. Hence it is challenging to use these AI algorithms in practice as these false positives add workload on doctors (for manual examination) and also can lead to unnecessary patient anxiety.

mCardia and DeepAware hence can substantially reduce the physician’s workload of manually reviewing the false positives and facilitate long-term ambulatory monitoring for early detection of arrythmia in patients natural setting. Also, the DeepAware holds the potential to get integrated into wearabletechnology like watches etc., for more accurate detection of heart rhythm related serious life-threatening condition.

CONGRATULATIONS TO DR KAMAL SHARMA

(IMAGE COURTESY TOI)

HELLO ALL,

IF YOU HAVE SOMETHING: PRODUCT,SOLUTION OR DEVICE OR TECHNOLOGY AND  YOU WANT  TO SHARE WITH HEALTHCARE PROFESSIONAL ,PLEASE GET IN TOUCH WITH US,INNOVATION AND NEW THINGS ARE CLOSE TO OUR HEARTS,WE ENCOURAGE NEW TALENTS AND INNOVATION.