Machine Learning and AI Underpin Predictive Analytics to Achieve Clinical Breakthroughs

Categories: General Healthcare Life Sciences Machine Learning Natural Language Processing

The practice of medicine is not only a science, it is also an art.  With it, difficult situations will arise requiring insightful judgments made by well-trained physicians who can tailor their approach to the needs of patients.  As such, we are witnessing a revolution in the healthcare industry, in which there is now an opportunity to employ a new model of improved, personalized, evidence and data-driven clinical care.

Physicians have long favored summarizing their patient encounters in freeform clinical notes which become part of the patient’s Electronic Health Record (EHR). Despite advances made in EHRs of late, they, unfortunately, do not provide advanced analytics or intelligent search for that matter. Providers trying to gain insight are challenged because up to 80% of all healthcare data is unstructured and combines EHR, clinical notes, genomics data, and more. For those asking big questions, in the case of healthcare, an incredible amount of insight remains hidden away in troves of clinical notes, EHR data, medical images, and omics data.

To arrive at quality data, organizations are spending significant levels of effort on data integration, visualization, and deployment activities.  Additionally, organizations are increasingly restrained due to budgetary constraints and having limited data sciences resources.

It is fair to say that healthcare faces many challenges, including developing, deploying, and integrating machine learning and artificial intelligence (AI) into clinical workflow and care delivery. Security and compliance must be met, first and foremost. Analyzing a variety of data not limited to, clinical and biological data will be necessary to gain valuable insight. Having the proper infrastructure with the required storage and processing capacity will be expected in order to efficiently design, train, execute, and deploy machine learning and AI solutions.

Together in tandem with MetiStream, a healthcare analytics software company, Cloudera addresses many of these challenges.  We recently announced the availability of MetiStream Ember on top of Cloudera, which offers an end-to-end interactive analytics platform specifically for the healthcare and life sciences industries.  Now organizations can reap all the benefits of having an enterprise data lake, in addition to an advanced analytics solution enabling them to put machine learning and AI into action at massive scale to improve health outcomes for individuals and entire populations alike.

The combined solution consists of Cloudera Enterprise Data Hub, Cloudera Data Science Workbench, and MetiStream Ember software with options for both cloud and SaaS deployments.  The joint solution provides the following core capabilities:

  • Automated natural language processing (NLP) of unstructured clinical notes, in order to store as structured, reportable data that can be rapidly searched and analyzed intelligently.
  • Real-time and batch processing from any EHR system including multi-year historical loads. Ember Clinical Notes pipeline annotates free text to discern clinical terms and normalizes these terms to well-known medical ontology codes, most notably UMLS CUI, SNOMED-CT, ICD-9, ICD-10, and RxNorm
  • Turn-key enterprise data integration (EDI) accepting both batch and real-time integrations for a variety of data sources and achieving data ingestion for EHR, omics, medical images, claims, drug data, vast sources of Real World Data (RWD), and social determinants of health (SDOH) data
  • Next generation Fast Healthcare Interoperability Resource (FHIR) standard for both data ingest and as a universal schema. Ember exploits FHIR beyond data exchange to empower interoperable analytics
  • Out-of-the-box advance analytics capabilities to eliminate 50-60% of costly ETL, data integration, visualization, and implementation.  
  • Actionable healthcare analytics  that allows organizations to conduct real-time “what if scenarios” against predictive models
  • The Ember platform is web and mobile-enabled to enhance collaboration and interaction between health providers and patients
  • Lastly, the solution supports HIPAA policies and procedures, including security capabilities ensuring user authentication, authorization, and full encryption of datasets, both on-premise, and in the cloud

Many of the turnkey analytic capabilities in combination with the underlying foundation of Cloudera are powerful in driving significant healthcare transformation. The combined solution is the right starting point for introducing machine learning and AI into clinical workflow and care delivery, enabling enterprises to take advantage of complete descriptive, predictive, and prescriptive analytic capabilities.  Start with automated natural language processing of clinical notes, to joining other quality clinical and omics data, and train and execute models in order to identify patient risk, predict patient decline, improve service quality, and utilize interactive analytics to better engage patients and improve their health outcomes.

Industry leaders have already recognized significant cost and time savings as well as increased capacity for improved patient care with machine learning through their work with Cloudera and MetiStream.  Please view our announcement and solutions gallery page on Healthcare Analytics for additional customer and solution details.

Author: Ryan Swenson, Global Leader, Cloudera Healthcare and Life Sciences

facebooktwittergoogle_pluslinkedinmail

Leave a Reply