Big Data and AI implementation in the Healthcare
Our contribution to cognitive Healthcarey
Healthcare payers and providers traditionally use multiple databases to create reports and gain insights which creates a challenge to consolidate data on-demand. This can be addressed by building a data lake using Hadoop to integrate data across multiple sources and in multiple formats like DICOM image to support dynamic reporting
Real time access to healthcare data for quick decision making
The chargemaster analytics solution quantifies and verifies whether payments to providers are within permissible limits as laid down by Contract Language and Charge Description Master. The solution uses advanced statistical and machine learning algorithms – both in the discrete and continuous domains – to detect billing amount increases with high confidence
Detects an increase in billing amount
Claim Adjudication helps payers to address the issues of over-payment and pay as per contract guidelines. Our solution uses natural language processing to understand the context of the item description and classify each item description into a charge type (e.g. laboratory changes, monitoring charges etc.) using Machine Learning algorithms to accurately categorize diverse healthcare costs in itemized billing and automate the adjudication process
40% increase in assessment speed for claim adjudication
Underwriting group insurance policies is a tough task as multiple members have multiple factors for premium determination. This solution involves improving the pricing and customer service for its group insurance customers by predicting medical cost through a statistical model using input data like Claims data, Enrolment Data, Prescription Data and Member Data
Improves cost prediction accuracy by 3% above the traditional underwriting model
Medical screenings are conducted for early detection of potential health disorders and diseases to ensure risk mitigation. The solution, utilizing Machine Learning algorithms, pinpoints eligible individuals who have not availed medical screening facilities. Business rules, adhering to HEDIS measures and guidelines have been incorporated into the algorithms
Income from Quality Performance Measure (QPM) incentives awarded on a PMPM (Per Member Per Month) basis
Manually reading scanned physician notes is a non-scalable and inefficient business model to determine risk conditions. The solution helps in reading the scanned prescription notes using Computer Vision and Deep Learning Models and categorizing the insights into disease, procedure, body organ, and drug-using a medical ontology
42% reduction in time to analyze the notes
Analyzing patient sentiments from feedback received is imperative for providers to improve the overall patient experience. Machine learning techniques are being used to automatically categorize patient feedback as either positive, neutral or negative and relate them to separate business aspects
Improves patient experience by providing accurate assessment of patient opinions about different aspects of the hospitals
Providers face huge challenges in retrieving and analyzing massive volumes of ECG image metadata from fragmented data storage. The solution by Abzooba migrated data to an integrated data lake based on Hadoop and generated relevant metadata for each ECG image using Machine Learning and image processing components
Efficiency of processing huge volumes of image metadata at scale has increased by almost 70%
