Since Big Data has proven its usability in retail, marketing, and other areas, healthcare managers are now thinking about how to reap the benefits of this technology for their own problems. Artificial intelligence in the form of Natural Language Processing (NLP) can improve critical aspects of the patient-doctor relationship and can even go beyond this, simplifying the process of insurance payment.
The expected advancement comes from making the clinical documentation more accessible through automatic indexing, thus adding search-ability. Another growth direction is the automatic voice-to-text feature which will enable the creation of automated digital records while allowing medical staff to focus their attention on patients instead of writing. This is solving the problem that more than 83% of physicians have reported: burnout.
Virtual Nurses and Assistants
AI-powered personal assistants like Siri, Cortana or Alexa are already part of our daily lives, and we’re used to asking them for information, interacting with them in various ways and relying on their help. However, these are general-purpose tools. To perform in the medical field, it would be useful to have similar, medical-grade assistants.
These virtual nurses could take notes and help doctors or even their human counterparts. For example, a smart system could take notes while a human nurse assesses a patient’s vital signs through voice activation, thus freeing up the nurse’s hands.
Also, the NLP-capable system can check on the patient to make sure they have taken their medicine, re-order some drugs as it sees that stocks are declining. Chatbots can also take on some of the work currently performed by office assistants, nurses or even doctors for smaller practices. These tools can also create semi-automatically generated e-mails, handling scheduling or calendar management.
Self-Help for Patients
The rise of technology has made most of us very independent and self-reliant. We don’t ask for directions anymore; we don’t need outside help. However, this attitude only works fine with non-critical aspects of our lives. Self-diagnosis using Google is incredibly dangerous. However, there are ways to create technology to improve our health in other ways.
Most clinics give their patients the option to access their data online, for example, seeing their results on portals or e-mails. This is an advantage for patients as most of them can’t wait to learn if they are okay or not. Text analysis of their EHR could help them translate specialized medical terms into simple information about the state of their health. This would contribute to the right decisions regarding lifestyle changes and active management of their conditions. Also, it would mean that fewer visits to the doctor’s office would be required, thus helping those who live in remote areas or have limited mobility.
Through smart tools like these, you could increase the average patient’s knowledge of their health by helping them understand specialized language by simply asking the system, much like you ask Siri. Such a tool can make a huge difference for the peace of mind of the patient and save them and the doctor time which would have been used on such explanations.
The next step is to create a medical lay-man dictionary of terms. Such a tool can be put together by applying NLP to medical records and could become the equivalent of Google Translate for medical documents.
Better Care through Information
As NLP turns text into actionable information, this opens up new opportunities related to unstructured data lying around in medical records or other annotations doctors have made in their observation sheets. Each patient has a unique story, development, and individual issues. If these individual data points could be gathered, centralized and structured, new patterns of disease and cures would emerge.
Free text has this kind of limitation compared to structured data which is suitable for statistical analysis. Due to the advancements of text analysis algorithms, experts from InData Labs highlight that better care will be available in a few different ways.
First, processed textual observations can be analyzed and used as exclusion criteria. This is important in the case of drug cross-interactions, allergies and the existence of secondary conditions. If such a rule is met, the patient will not be given a certain drug which could cause them more harm than good, thus preventing additional complications
Next, the information can be included in quantifiable structures and become the foundation for reports. Qualitative information such as “heavy smoker” or “occasional drinker” could be understood by the software and coded appropriately, in a similar way this would be made on a scaled survey. Subsequently, this could be used to compute various risk scores.
Lastly, the information existing in a narrative format in a patient’s EHR can be used to qualify the individual as being representative of a particular group. This could be helpful in identifying the right candidate for clinical trials, the best options for revolutionary treatments or those who are at the most significant risk.
Current State and Future Developments
Currently, although these technologies have tremendous potential, they are underused, as Gartner showed in a study from 2016, there were under 5% of health systems benefiting from these advancements. The remaining healthcare providers rely on coded data according to existing healthcare standards.
The problem with this approach is that it doesn’t cross reference conditions and treatments, thus having a unidimensional approach. Once text analytics become the norm, we can expect that patients’ EHR will become comprehensive data sources for multi-layer analysis. This will lead to a much broader understanding of patient populations, disease development and possible co-morbidities or treatment interactions.
Another significant advantage of using these systems is their scalability and replicability. While currently there are only a handful of specialists who have competences in diagnosing some diseases, with such a system backed up by loads of data, identifying a certain medical condition or the high risk of developing it can be done by simply feeding the EHR into a software.
We can expect that the text analysis powered by NLP to become a more prominent and widespread tool in the future. In a decade, we might have apps which replace the majority of our current trips to the doctor’s office.