Healthcare CTOs must consider Natural Language Processing

Natural Language Processing, a form of Artificial Intelligence (AI) is a term that has gained serious momentum in the past few years. Essentially, it’s the computational decomposition of language, putting ‘human’ language into a form that computers can understand. 

Just recently, a new report from IBM found that NLP has emerged as the type of AI that appeals the most to organisations that want to automate processes. Clearly many businesses are beginning to understand its power. That same report found that 52% of global IT professionals report that their company is using or considering using NLP solutions to improve customer experience, with 43% using it to increase cost efficiency. 

When we look at NLP from a healthcare lens, it can act as a transformative tool. NLP provides the ability to improve areas of the business such as customer service, assist with operations and R&D, provide summarisation of media and text, as well as hiring and HR (avoiding gender language biases in job specs.) 

However, as healthcare is typically a very human-to-human industry, it’s only just starting to explore NLP’s uses, but they should. The opportunity is endless, and, when we take into consideration that the healthcare sector is typically lagging behind in digital transformation, it’s an option that healthcare CTOs shouldn’t overlook.

A linguistics maze 

We as humans will always structure language in certain ways to communicate with others. Today, there are roughly 6,500 languages and these languages often vary from cultures within the same languages, as well as informal and formal tones. 

Luckily, there’s a lot of power in NLP that helps take this information and turn it into something that computers can understand, regardless of language. Ultimately, language is a complex maze, but technology can navigate this, enabling healthcare professionals to better interact with patients, while giving back employees more time to care for patients.
 
Currently, NLP use cases are quite broad - whether it’s email or text classification, machine translation, virtual agents for customer service, call center automation or analysis and search of complex documents – all can act as transformative tools for healthcare businesses. 

Life-saving capabilities 

Algorithms powered by NLP are able to recognise and answer patients' questions online, such as the World Health Organisation’s site search. It can also recognise doctors’ questions and patients’ answers by extracting key points and putting them into electronic health records within seconds, thus reducing doctor’s workload and enabling them to give patients more one-to-one personal care.

Although it may seem that in the healthcare industry today, the main use of NLP is its ability to receive information and process, its benefits can even go beyond that. NLP can provide a greater degree of personalisation of care. 

NLP has the ability to offer much more empathetic treatment. Feedback surveys can utilise sentiment analysis by understanding the different language used. For example, for patients with HIV, doctors will often focus on the efficacy of drugs and blood counts, but there can be some awful side effects to the drugs. With NLP, if someone is recording how they’re feeling, getting sick, going to the toilet etc. then this can become a conversation between the patient and the clinician. NLP opens up the conversation that can go on to benefit the patient.

Yet it doesn’t stop there. NLP combined with machine learning (ML) in healthcare can help physicians make better decisions and aid clinicians in checking symptoms and diagnosis. In fact, one study in 2018 used NLP to predict suicide attempts by monitoring social media. The system had a 70% prediction rate with only a 10% false positive rate.

And while the statistics appear quite remarkable, the process behind it is simple; it breaks language down into data, making sense of it.

Unleashing the value of data

In the past, we wouldn’t have thought of the number of steps we take or our heartrates as data, but that’s changing thanks to activity trackers. And the healthcare industry is now able to make use of this data. For instance, Vitality is able to provide members with deals and offers based on their activity level, so why not use language data to also benefits patients?

Everything we see, we write down, and until recently doctors wrote all their notes by hand and then input very precise information into a computer. Now, doctors can tell a system to capture all the information they’re inputting, even if it has a very rigid input of data that may be restrictive. 

NLP can help extract all the relevant information like notes, diagnoses, patients records, typing up all the handwriting – all the time consuming tasks. There are already several health start-ups offering this to healthcare organisations with transcription technologies that capture text and use a remote human transcriber to edit the automated text and produce a “structured” set of notes from patient visits.

And just like other industries, healthcare can go from the ingestion of data – who the patient is, their health background, determining their past – to classification which transforms data into something useful.

What this means is that there’s strong potential to start taking pretext data to predictive, giving healthcare professionals an understanding of what might happen in the future and patient treatments. This isn’t a structured concept because humans are difficult to predict, but NLP can help understand the most likely future issues and include example health profiles. That means, with preconceived information, NLP can create a more precise diagnosis as it has more accurate, recent data.

NLP must combine with human intelligence 

Clearly, the benefits of NLP are endless. That’s not, however, to say that the healthcare industry should just be applying NLP without understanding how clinicians will use it. Without understanding the inputs and outputs then it just isn’t useful.

Like any technology, we must combine its power with human capabilities. As many healthcare AI companies are run by white Anglo-Saxon males, understanding cultural nuance and inclusivity is crucial to improve patient benefits.

NLP cannot be a substitute for human intelligence. It needs a kind of bionic-thinking, the synergy between human and machine, to allow NLP to really shine giving it a central role in powering innovation, efficiency, resilience, and most of all, allow healthcare professionals the opportunity to provide a greater level of care for patients.