Debilitating Disease - plagued by your own data swamp
Amyloid plaques and neurofibrillary tangles- that is what was found on the first patient that the German psychiatrist, Dr. Alois Alzheimer diagnosed in 1906 from a woman who had died of an “unusual mental illness”. The patient was plagued with memory loss, language difficulties, and erratic behavior. The autopsy revealed abnormal clumps and tangled bundles of fibers, which left little imagination of why she was acting so strangely- her mind had turned into her own swamp. These now are known to be stereotypical findings of the disease that took on Alois’s name- Alzheimer’s disease. With advanced imaging techniques, we have learned that there is actually a loss of connections between nerve cells in the brain. This information is described in detail within pathology reports usually represented as attachments within the Electronic Health Record (EHR). Alzheimer’s disease and many other diseases are debilitating for patients and loved ones; therefore, it is vital to capture such information to help diagnose as early as possible. With all the advancements in imaging and medicine, data & AI have stepped into a leading role in identifying early onset and treatments for disease.
A focus on Precision Medicine
Linguamatics is working with Washington University School of Medicine as part of the school’s effort to advance personalized medicine research. The school’s initial focus is on Alzheimer’s disease, breast cancer, lung cancer, diabetes, and obesity – diseases that affect millions of people worldwide. But in order to take a successful dive into this research venture, it’s imperative to know what you are dealing with within that database, before you take a swim. You need to compare apples to apples, or rather gather imperative text-based information and normalize the structured and unstructured data to identify and extract relevant clinical and biomolecular phenotypes. Information that just wouldn’t be available without AI tools like Natural Language Processing (NLP). Such as: “[x dosage] of [medication name] was taken at [this interval]”, and “patient’s [family member(s)] had dementia”. Imperative information that leads to innovative research isn’t limited to structured data alone. Clinicians spend hours documenting for a reason - this information is important for patient care. And this information is well beyond the limited data that is relatively easy to extract from structured clinical fields. Linguamatics NLP is utilized so this important documentation makes it into the dataset so it can be available to researchers for future use.
"Key information needed for a full understanding of a patient's condition and treatment is available only within the free text EHR. NLP is now in widespread use to provide access to this information. NLP captures the different ways information can be expressed, and normalizes the data, including the use of standard concepts identifiers for diseases, medications, phenotypes, genes, or gene mutations" explains David Milward, Senior Director NLP Technology, Linguamatics, an IQVIA company. “Data extracted can also be linked with scientific knowledge that has similarly been normalized using NLP from free text articles."
"As part of our effort to advance personalized medicine research, it is critical to understand the range of health data collected about diagnoses, treatments and treatment outcomes,” said Philip Payne, Associate Dean for Health Informatics and Data Science, Washington University. “But gathering usable data from de-identified unstructured medical records in a way that can be incorporated easily into personalized medicine research is a challenge. We’re collaborating with Linguamatics on this effort and will be evaluating its platform."
How does natural language processing help healthcare?
Simply stated…a better picture means researchers find better answers, which then translates to better patient care. It is innovations and partnerships such as these that are necessary for the future of medicine. It’s a much better future if we can work with the whole picture, rather than just pieces of the puzzle.
On-demand webinar: Natural Language Processing, Rare Disease and Precision Medicine