>Business >Unstructured data – coming to the healthcare party

Unstructured data – coming to the healthcare party

The United States healthcare field now generates an approximate 1.2 billion clinical care documents annually. In those documentations is an astounding degree of detail on an individual patient’s care plan, medicinal history and cumulative health. So why is this variant of data, not often touted, critical? 

When put under analysis, this comprehensive patient data has the potential to enhance healthcare outcomes. The only issue is that approximately 8/10ths of that data is of the unstructured variety. 

Longitudinal data – data that consists all healthcare encounters of a patient or member over a period of care of time, is really critical to finding out insights about illness, comprehending the patient’s health trajectory and for the optimization of treatments. Some if this information is considered unstructured – implying that it has been put into a standard format and is easily accessible and shared via a database. Unstructured data, which contributes to the dominant majority of longitudinal data sets, is much more difficult to standardize and thus more difficult to access, share, and analyse. 

Unstructured data consists of everything ranging from email communications, practitioner clinical notes, patient phone call transcriptions, comment cards, wellness diary entries, lab reports and documentation, socioeconomic data, patient preferences, key lifestyle factors to X-ray imagery and faxed lab reports and documentation. It consists of other interactions members have with the system, like prescriptions, lab values in addition to all of the data that is obtained through call centre and care manager interactions. Bringing this together with unstructured data with available structured data with available structured data furnishes a more comprehensive picture of a patient’s cumulative health profile. 

A last piece of the patient data puzzle comes from the five critical spheres of social determinants of health (SDoH) which the Office of Disease Prevention and Health Promotion defines as economic stability, education, social and communal context, health and health care, and neighbourhood and built environment. The National Academy of Medicine states SDoH accounts for 8-9/10ths of the modifiable contributors to health outcomes for a particular population, so it’s critical to integrate these factors when developing a holistic perspective of the patient. 

Avenues from unstructured data 

Bringing together diverse structured and unstructured data furnishes massive opportunities for the healthcare field to obtain the highest level of unique insights possible and influence patient care. Preliminary attempts to gather data from unstructured data in healthcare concentrated on simplistic analysis leveraging keywords that could provide insights not available in the structured data of EMRs. These notes assisted in determination of the level of severity that might not be immediately apparent from clinical test outcomes. 

However, there is a lot more that can be done. The healthcare field ought to be modelling its efforts consisting of unstructured data after other industries that are already reaping the benefits of this opulent source of data. More than 50% of the data leveraged by retail, travel/hospitality, energy, insurance, consumer goods, and financial services to obtain insights is unstructured – more than any other variant of data. The retail industry leverages this data to comprehend client behaviour and engage clients in a targeted manner. The same can be performed for patients. 

The challenge is furnishing a level of accuracy that will facilitate scientists, care managers, and physicians to access this data on a real-time basis in every scenario so they can swiftly act upon it. Artificial intelligence (AI) and machine learning are critical to tapping into the wealth of unstructured data to unveil insights that may have in other instances, been victim of oversight. 

Utilizing AI to leverage unstructured data 

Process of the terabytes of available unstructured data and converting them into something actionable is without question a challenge. The activity needs a massive amount of computational capacity and artificial intelligence capable enough to unveil the intricacy of this information to discern the signal from the noise and furnish insights that are unique and actionable.  

To get through the hype and decide the realistic possibilities of artificial intelligence technologies within healthcare, which includes unleashing the insights of unstructured data, the Office of the National Coordinator for Health IT (ONC) and the Agency for Healthcare Research and Quality (AHRQ) resorted to JASON – an autonomous unit of researchers, scientists, and academics that consistently provide advise to the federal government. 

Amongst its recommendations, JASON is compatible with capturing and utilizing unstructured data from mobile phones in addition to societal and environmental information. The effort drives home the commitment for the ONC and AHRQ to efficiently leverage this variant of data to enhance the quality and safety of patient care. 

After the data has been gathered, one possibility is to work with Natural Language Processing (NLP) algorithms that are particularly developed to take unstructured text and convert it into a structured format. There are various efforts in progress in converting EHR notes, practitioner notes, chat convos, and other text fields into usable data.  

The question then turns, after this fresh influx of information is formatted and available how do you leverage it? In a word or two, artificial intelligence. (AI) AI is beginning to revolutionize the healthcare domain ranging from how drugs are produced to how care is being furnished to detecting the ideal treatment for an individual patient on the basis of his or her biology, and prior medications that they have been prescribed. To efficiently utilize this growing healthcare data set, we require an artificial intelligence technology such as casual machine learning that is data agnostic and can convert data from the really granular genomic to the more systemic EHR data to identify the underlying reasons and impacts within the data. This is a very unique technology in contrast to mere scanning of available data, it is finding out insights we didn’t know that we weren’t aware of, prior. 

So what does this have in store for the patients? It implies that the data is bigger, opulent, and more accessible. It implies that the technology persists to progress and become smarter with the passage of time. And it also implies that the concept of precision medicine where patients receive treatments on the basis of their individual biology is no longer a distant dream, but a close by reality. 

Add Comment