Enhancing problem list reconciliation with Natural Language Processing (NLP): Improve patient care quality with population health text mining analytics
The shift from volume to value-based compensation is driving provider demand for better insights into the health of patient populations. Providers recognize that access to more complete patient data can enhance their ability to deliver cost-effective care and high quality outcomes. This is especially true for patients with multiple chronic conditions, who typically have more complicated care needs and higher hospital utilization rates.
Figure 1 High risk patients are frequently suffering from complex comorbidities
Typically, physicians refer to problem lists when assessing a patient’s health and evaluating treatment alternatives. Problem lists rely on coded disease states and offer a concise view of a patient’s medical issues. Unfortunately, these lists are often incomplete or out of date. Consider, for example, a patient who is referred to an orthopedic surgeon for a broken wrist. If the problem list only includes details of the wrist injury, the physician may not be immediately aware of underlying chronic conditions, such as diabetes, that could impact the best course of treatment and outcomes.
Critical patient health details are often hidden within unstructured free text. A patient’s HIV status might be indicated within a free text section of the history and physical, but not included on the coded problem list. For this reason, doctors must look beyond the problem list and manually review the non-structured sections of a patient’s chart to identify other relevant medical information. Manual chart reviews can be time-consuming, but are essential for obtaining a full picture of a patient’s health and identifying underlying health issues and comorbidities that can compound risk.
Fortunately, new technologies such as Natural Language Processing (NLP) can speed up the problem reconciliation process and help providers to identify critical details hidden within free-text sections of the chart. NLP tools can filter clinically relevant data from unstructured patient-related documentation; key information can then be easily extracted, allowing clinicians to assess what items to include on the problem list. Take discharge summaries as an example. When a patient is discharged, the discharge summary details all relevant information from their hospital stay. Using NLP, it’s possible to extract these diagnoses and cross reference them with the current problem list. Any required updates are flagged to the clinician for review as part of a reconciliation task in the EHR task list. NLP speeds up the reconciliation process and ensures a more accurate and complete problem list – which can significantly improve the delivery of care and enhance patients’ long-term health.
Want to learn more about mining unstructured patient data for successful population health management?
