Project Description
Developing Novel Computational Methods to Help Integrate Palliative Care in the Treatment of Advanced Heart Failure
Efforts to improve care for patients with life-limiting illness have been
stymied by our inability to access patient data with the granularity and
timeliness required for predictive analytics and outcomes research. This
project seeks to uncover the narrative text of medical notes using Natural
Language Processing (NLP) and machine learning to identify the full range of patients’
experiences and outcomes.
The increasing use of electronic health records combined with computational advances in NLP and machine
learning have the potential to transform the narrative of medical notes into
quantitative variables ready for use in statistical models. While underutilized
in medicine, these methods are ubiquitous in business where NLP has
revolutionized fraud detection, product recommendation, speech recognition, and
customer segmentation. Similarly, NLP and machine learning have enormous
potential to transform clinical research by unlocking access to free text variables,
enabling complex pattern detection, thereby improving prediction accuracy. While
useful in all aspects of health care, these tools offer particular relevance to
palliative care, where the focus is on the qualitative experience of the
patient and family.
Advanced heart failure presents an excellent test case
for this approach because of its large patient population of more than 5.8
million in the United States, the challenge associated with predicting treatment
response and mortality, and the tremendous variability in quality of life over
the trajectory of illness. Using computational methods, we will extract free
text variables from medical notes describing the patient and family experience
over the illness trajectory of heart failure. We will specifically test the
value of the natural language of medical notes to predict one-year mortality in
heart failure. Our approach will expand the data sources available for
palliative care research and better define patient needs and the impact of
palliative care.
Bio
Charlotta Lindvall earned an MD and PhD (Medical
Genetics) from the Karolinska Institutet in Stockholm, Sweden, and was a
researcher in cancer genetics for 7 years. In 2010, she undertook clinical
training in the United States, completing residency in Internal Medicine and a
Harvard fellowship in Palliative Care. Dr. Lindvall is currently completing a
2-year clinical research fellowship in general medicine at the Massachusetts
General Hospital. She will join the Department of Psychosocial Oncology and
Palliative Care at the Dana-Farber Cancer Institute in July 2016. Dr.
Lindvall’s research interests include developing computational methods to harness complex clinical data
extracted from the electronic health record, with the goal of facilitating shared
decision-making between seriously ill adults and their health care providers.
Her research is supported by a collaboration with the MIT Computer Science and Artificial Intelligence
Laboratory.
Email: Charlotta_Lindvall@DFCI.harvard.edu