Authored and contributed to PharmaLeaders by Jason Bhan, MD
Accurate analysis of healthcare data is critical for research, delivery of care, treatment variability, and improving patient outcomes. The ability to identify patterns in real-world data (RWD) demands insights with faster speeds, greater specificity, and improved accuracy. And the challenges are increasing as more healthcare data is introduced. It takes a combination of data science, clinical expertise, and technology to extract clinically significant insights from siloed healthcare data. The next step is to make these insights available at scale. It’s a massive task, but technologies like machine learning (ML) can be leveraged to make the process more systematic, efficient, and accurate.
Importance and Challenges of Lab Data
Among all of the healthcare data accumulated along the patient journey, lab results are the most influential dataset driving more than 70 percent of clinical decisions.1 Yet this data is often the most complex and challenging to manage at scale.
At a time when 14 billion lab tests are ordered each year in the United States1 alone, the sheer volume of lab data is a Herculean task to collect, analyze, and interpret. In addition, lab data’s unstructured nature and lack of standardization across the industry make it particularly difficult to manage and apply.
Effectively identifying patterns in lab data is further complicated by the inconsistently formatted clinical notes that are typically added to lab results in a non-standardized way (i.e., manually). This arises because lab test results are intended to be read by a human for analysis of a single patient.
These challenges of volume, lack of structure, non-standardization, and storage and interpretation limitations leave many life sciences companies without the resources necessary to leverage this valuable data source effectively. Yet the ability to manage lab data at scale, seamlessly integrate it with other clinical data types, and generate consequential insights based on the full patient-level view of the data is a function that creates significant value for many use cases in the pharmaceutical space.
A workable solution requires an integrated infrastructure to consistently collect, store, monitor, and analyze data at an ever-accelerating pace. Companies that rely on identifying patterns in data to reveal clinical insights are beginning to realize an obvious truth: machine learning is rapidly altering and improving how data, especially lab data, is managed and applied.
Machine Learning Meets Laboratory Data
When companies understand the significant value to be gleaned from lab data and the common challenges it presents, they can find the solutions needed to effectively ingest, manage, and analyze the data. But how can the data be processed and interpreted to generate insights about the patient journey at scale? That’s where machine learning comes in.
Machine learning, a subset of artificial intelligence (AI), teaches computers to think in a similar manner to humans. The beauty of ML is that it learns over time, becoming better able to accurately analyze data. Ultimately, the goal is for the technology to provide data capture and analysis seamlessly to integrate with the expertise that clinicians and other healthcare providers bring to the patient journey. Once trained, machine learning models are able to identify complex patterns.
Recognizing the potential of ML, data scientists are partnering with clinicians to develop models that can effectively and efficiently address the challenges presented by lab data and identify clinical truths within the data to create real value. Because these models become increasingly accurate over time and can generate the initial interpretations and mapping that were historically derived from manual processes, ML is significantly shortening the time to insight.
Natural language processing (NLP), a subset of ML, plays an important role in dealing with the multi-dimensional challenges of interpreting lab data. NLP is the ability of a computer to understand, analyze, manipulate, and potentially generate human language. For example, we know that important information is frequently buried in the unstructured text fields of lab results. NLP can be used to extract this information and convert it into a valuable knowledge source. NLP capabilities are always evolving, but it is already being used to read pathology notes, translate healthcare acronyms, decipher typos, and learn and apply new terms. The information pulled from these unstructured clinicians’ notes adds clinical specificity to the patient-level data that has not been accessible at scale in the past.
Armed with these ML and NLP solutions, companies can ingest and interpret the nearly endless supply of healthcare data available in the ecosystem instead of being limited to normalized data sets from single sources. And while computer models are equipped to increase the speed and quantity of data ingestion and interpretation, the clinician remains a critical piece of the process, ensuring maximum value is derived from the data.
Going Beyond Lab Data
The real power of machine learning in this context comes from easing the integration of lab data with other patient data to provide the most complete view of the patient journey. ML can empower more timely and accurate data integration and analysis that can be applied to clinical decisions to improve patient outcomes. When these strategies are employed, insights from RWD can be rapidly extracted to support different analyses, such as patient journey analytics, outcomes studies, provider evaluation, and patient segmentation. This type of data harmonization, now becoming more readily available with ML, delivers actionable insights that empower pharmaceutical companies to positively impact patient outcomes through timely and meaningful intervention.
About the Author
Jason Bhan, MD, is a Family Physician and is co-founder and chief medical officer at Prognos Health. He is regarded as a national expert in the applications of technology to healthcare and medicine, a topic on which he speaks regularly at institutions and conferences, such as Health 2.0, mHealth, New York’s eHealth Collaborative, and Health Datapalooza. He also has done extensive strategy consulting with pharmaceutical companies. From 2007-2010, Dr. Bhan worked with Clinovations and managed several large hospital system EHR implementations, outcomes measurements and data analysis. Dr. Bhan obtained his Doctor of Medicine at the University of Miami School of Medicine and he is board certified in Family Medicine.
References:
- Centers for Disease Control and Prevention. https://www.cdc.gov/csels/dls/strengthening-clinical-labs.html