Although human languages follow specific structures and patterns, they pose challenges for machines. In the early days, low-level machine languages that used binary codes (0 and 1) were commonly used for human-to-machine communications. Later, high-level languages such as C++ were developed. But it is only in recent times that machines are able to interact with human languages. With the advent of artificial intelligence (AI), machines can process human languages.
In the early days of AI in linguistics, its use was envisioned primarily for translation services (e.g. English to Spanish). But soon enough other language-related applications were developed. Machines could perform simple tasks such as autocorrection or email response suggestions with accuracy. Natural language processing (NLP) is the branch of AI that enables machines to read, comprehend, and work with human languages. At present, machines can interpret and work with spoken language or written text. This makes NLP well-suited for solving various problems related to big data in the healthcare domain.
NLP applications can help in performing various repetitive administrative tasks that may otherwise require trained personnel. For instance, in hospital settings, NLP can be employed for information extraction and document categorization. NLP applications can read doctor’s notes and extract relevant information to accurately assign billing codes. This will help in pre-approvals or timely authorizations of treatments, which ultimately reduces the burden of illness.
Accountable care organizations can use NLP to improve the patient journey. Chatbots or virtual assistants can provide information whenever requested by the patient and thus reduce the anxiety about hospital visits or medical procedures. This also allows patients to plan other aspects of their life such as caregiver schedules, which ultimately creates a positive patient experience. Additionally, a patient could ask for information about activities they can do or need to avoid after undergoing surgery. Chatbots can discuss the options and be useful in improving the quality of life for patients. NLP can also be utilized to study social media posts for understanding patient journeys.
Manual reviews of medical records are time-consuming. NLP applications help in reducing the time required for manual expert review of unstructured data such as electronic health records (EHR). Doctors and other healthcare professionals are already burdened with paperwork. Additionally, humans are prone to errors and omissions as fatigue sets in. NLP applications have the ability to review, analyze and sort the EHR into meaningful data and meaningful insights. In the case of clinical trials, safety reviewers read narratives from adverse events, medical notes, and also medical literature. The regulatory authorities that provide oversight - the US Food and Drug Administration (FDA) and the Centers for Disease Control and Prevention (CDC) consider NLP as a possible solution for unstructured data.
Real world patient reported data often has missing data and NLP applications can be used to find this missing data. A study done in the UK used NLP for text mining data from free text EHR. The occupation of a patient is usually recorded in a structured field as categorical or numerical data. In mental illnesses, occupation in structured fields is often missing as the patient may be unemployed or is still a student. However, a doctor or nurse may have written it as free text in their medical notes. Given that mental illnesses and occupations are correlated, researchers need data about the occupation. The research team developed NLP applications for data mining and identified occupations for a majority of patients. Identifying occupations helps in designing policies for occupational placement and providing increased support for patients.
Pharmaceutical companies have focused on the use of NLP for clinical trials. A recent study surveyed fifteen pharmaceutical companies to study their understanding of pharmacovigilance initiatives. The survey report has identified areas such as ‘language translations, case verification, in-line quality control, prioritization/triage, data entry, alerts for cases of interest, workflow management, and monitoring where automation technologies’, including NLP can be effectively used. NLP applications can further be useful in qualitative studies where data can be extracted from focus groups, interviews, or questionnaires.
NLP applications help in extracting meaningful insights from the data. RLS has a trained data team that has the experience and expertise to work with available data. Real Life Sciences Technology Platform offers powerful tools to aggregate data, analyze it using standardized medical terminology, and provide answers to interesting research questions. Recently, a study analyzed social media narratives from patients and caregivers with Alzheimer’s disease to understand the disease burden. Using this novel method of social listening, narratives were categorized into the ‘Social, Physical, Emotional, Cognitive, and Role Activity (SPEC-R) framework. The study revealed current gaps in data captured by clinical instruments and advanced our understanding of the disease burden. This organizational framework of SPEC-R can be applied to study other diseases such as arthritis, Parkinson’s Disease, and trauma.