Effective analytics in healthcare

By Jivendu Biswas
08:45 AM

In the world of modern medicine, it takes approximately seventeen years of original clinical research to be integrated into the day-to-day practice of medicine. With all this knowledge, asking an individual doctor to rely on his memory is like asking travel agents to memorize all airline schedules.

But let’s fast forward into the future. Imagine this:

  • Triage being aided by an Intelligent Knowledge Application.
  • All the medical knowledge in the world being available to a practitioner during a patient encounter.
  • The right care being given to the right patient at the right time – all the time.

The healthcare ecosystem is a complex one.

A disease has symptoms, which has a diagnosis code, which has a set of related sequential investigations followed by a secondary/tertiary diagnosis, which has associated treatments which differ as per each patient’s preexisting medical conditions and genetics. Associated treatments can be in the form of further follow-up diagnostics, medical procedures, durable medical equipment, counseling, acute care, long term care, and other forms of treatment.

This vast and growing medical information that represents increasingly complex relationships and dependencies presents a seemingly impossible situation. Most of the current analytics implementations are in silos and hence fail to capture the holistic picture, which at times lead to erroneous decisions and actions.

How do we assimilate, organize and present this knowledge at the right time, for the right patient, in the right situation, all of the time? Furthermore, as medical knowledge increases and therapies are discovered, the database continues to grow and grow. 

After much research, analytics professionals and knowledge engineers have found that Semantic Networks can be leveraged to simplify this complex information store and provide practitioners with the information needed to make more effective diagnosis and treatment possible.

The UMLS (Unified Medical Language System), developed and maintained by the United States National Library of Medicine, is an instantiation of Semantic Networks created specifically for healthcare.

The UMLS is a compendium of many controlled vocabularies in the biomedical sciences. It provides a mapping structure among these vocabularies and thus allows one to translate among the various terminology systems; it may also be viewed as a comprehensive thesaurus and ontology of biomedical concepts. UMLS further provides facilities for natural language processing and is intended to be used mainly by developers of systems in medical informatics.

Be it ICDs or DRGs, CPTs and their relationships, UMLS has all the knowledge, is updated quarterly and can be consumed for free.

UMLS is exposed to other systems using Application Programming Interfaces. These APIs are open to the world for building smart systems based on the knowledge stored in it.

The future healthcare information ecosystem can be around an analytics core that would ensure tight collaboration amongst all information silos and functional systems. UMLS-like knowledge hubs would be one of the levers. The analytics core would comprise of CDSS, NLP and associated rules and heuristics, which would analyze the patient knowledge, medical knowledge and compliance knowledge to provide appropriate responses to queries from various stakeholders including clinicians, administrators and others involved in the care delivery process, as depicted below.

Here’s a scenario: A patient comes into the ED. The system is fed with his symptoms. It collaborates symptoms with the patient health record, current seasonal information, etc., to predict a probable diagnosis and its severity. It suggests the appropriate wait time based on compliance data and severity, and allocates resources and alerts providers based on their availability, location and expertise, along with suggested investigations. It also recommends external resources if the case can’t be handled internally in that facility, and informs the other facility and transport/ambulance.

For analytics to be effective, we have to make decisions and take actions based on integrated, relevant and timely information in a holistic manner. With the right information, at the right time, for the right person, we will be able to make a dramatic positive impact on our ability to provide better healthcare outcomes with lower costs and improved patient satisfaction, with highest levels of compliance.