Blog from September, 2021

Hannah Nelson


Machine learning systems can mitigate burden and boost EHR usability for disease phenotyping to support clinical research, according to a new study.


Machine learning systems can aid EHR usability and cut burden for disease phenotyping to support clinical research, according to a recent Mount Sinai study published in the journal Patterns.

The machine learning-based algorithm diagnosed patients as accurately as the standard set of disease phenotyping algorithms for conditions like dementia, sickle cell anemia, and multiple sclerosis.

“There continues to be an explosion in the amount and types of data electronically stored in a patient’s medical record,” Benjamin S. Glicksberg, PhD, a senior author of the study, said in a press release. “Disentangling this complex web of data can be highly burdensome, thus slowing advancements in clinical research.”

“In this study, we created a new method for mining data from electronic health records with machine learning that is faster and less labor intensive than the industry standard,” continued Glicksberg, an assistant professor of genetics and genomic sciences and a member of the Hasso Plattner Institute for Digital Health at Mount Sinai (HPIMS).

Clinical research scientists currently use a standard set of disease phenotyping algorithms managed by a system called the Phenotype Knowledgebase (PheKB).

The study authors noted that while effective, implementing a PheKB algorithm on a new dataset is time-consuming as it requires variably formatted data, as well as specific laboratory or clinical information.

PheKB algorithms also have limited scalability since they are curated based on expert knowledge for one disease at a time, the researchers explained.

Only 46 diseases or syndromes are represented by public PheKB algorithms as of July 2020.

To develop a new algorithm for a disease, researchers must manually go through EHR data looking for certain data that is associated with the disease and then program an algorithm to identify patients with those disease-specific pieces of data.

The Mount Sinai researchers automated the disease phenotyping process through machine learning in an effort to save clinical researchers time and effort.

The researcher teams’ new method, Phe2vec, was based on studies they had already conducted.

“Previously, we showed that unsupervised machine learning could be a highly efficient and effective strategy for mining electronic health records,” explained Riccardo Miotto, PhD, a former assistant professor at the HPIMS and a senior author of the study.

“The potential advantage of our approach is that it learns representations of diseases from the data itself,” Miotto continued. “Therefore, the machine does much of the work experts would normally do to define the combination of data elements from health records that best describes a particular disease.”

Glicksberg noted that the study’s promising results suggest the algorithm could be used for large-scale phenotyping of diseases in EHR data.

“With further testing and refinement, we hope that it could be used to automate many of the initial steps of clinical informatics research, thus allowing scientists to focus their efforts on downstream analyses like predictive modeling,” he said. “We hope that this will be a valuable tool that will facilitate further, and less biased, research in clinical informatics.”

The study authors said that they plan to analyze how phenotypes change over time. They also plan to embed other kinds of data, such as genetics and clinical imaging, into the framework for refined disease phenotyping.

Additionally, they intend to explore the use of the system to create reliable disease-specific control cohorts for observational studies.






Hannah Nelson


Four of six travel intensive care unit nurses hired by a California hospital to address the ongoing COVID-19 surge quit due to poor EHR use training.


Four travel nurses hired by a California hospital to help address the ongoing surge of COVID-19 strains quit days after they were hired due to inadequate EHR use training and onboarding issues, according to reporting from the Times Standard.

Providence St. Joseph Hospital in Eureka, California brought on eight new traveling caregivers last week—six intensive care unit RNs and two respiratory therapists—according to a Wednesday press release. Four out of the six nurses quit the very next day, according to the Times Standard.

Ian Seldon, a spokesperson with the California Nurses Association, told the news outlet that the nurses left St. Joseph Hospital due to inadequate EHR training.

“Apparently, the travelers were met without necessary resources, including access to the unit’s electronic charting system and were immediately handed full patient assignments with little in the way of orientation,” Seldon noted. “So, four out of the six (travel nurses) quit.”

“In the words of one of them, the travelers were ‘thrown to the wolves’ and with all the opportunities available to travelers these days, they just didn’t come back,” Seldon explained.

Roberta Luskin-Hawk, MD, chief executive for Providence in Humboldt County, told the news outlet that the nurses’ departure was “an unfortunate and unique circumstance.”

“Some of the travelers who came to us through our request to the Medical Health Operational Area Coordinator did not stay at our hospitals,” she said. “The primary reason was that they were not familiar with our electronic medical record system — a system that is used by many hospitals.”

“Additionally, there were issues with the onboarding of these caregivers which created a challenge for them acclimating to our hospital,” she continued.

Luskin-Hawk said that Providence would continue to work with the Medical Health Operational Area Coordinator to find additional staff for St. Joseph Hospital as well as Redwood Memorial Hospital in Fortuna.

“We will continue, as we have throughout the pandemic, to aggressively seek additional resources focused on supporting our caregivers as they respond to the large number of patients requiring hospital services as part of this COVID surge while caring for our community’s important health care needs from open-heart surgery and trauma care to cancer care,” she said.

Luskin-Hawk also noted that the healthcare organization would be transitioning to a more popular EHR system to enhance care delivery across the health system.

“In addition to meeting the immediate needs of our communities, we are excited to be transitioning to a more widely used electronic medical record system in the coming weeks and will continue to work on additional projects that will enhance our health care delivery system over the near term and for years to come,” Luskin-Hawk told the Times Standard.

Effective EHR training programs may be the key to clinician satisfaction, according to a recent KLAS survey.

Researchers recommended healthcare industry stakeholders implement standards to ensure clinicians across health systems receive high-quality EHR training. They recommended at least four hours of EHR training to improve EHR satisfaction throughout the industry.

“Organizations requiring less than 4 hours of education for new providers appear to be creating a frustrating experience for their clinicians,” wrote the KLAS researchers. “These organizations have lower training satisfaction, lower self-reported proficiency, and are less likely to report that their EHR enables them to deliver quality care.”

Investing in EHR training may make the user more proficient at navigating the EHR, learning the intricacies of the platform, and it could potentially reduce the chances of clinician burnout in the future.  

“For EHR software to revolutionize health care, both the software and the use of that complicated software need to progress in parallel,” the research team concluded.





Hannah Nelson


API EHR integrations can support patient centered care by promoting patient reported outcomes data sharing with primary care providers.


Application programming interface (API) EHR integrations can support patient reported outcomes data sharing with primary care providers, according to a study published in JAMIA.

In a prior feasibility study, researchers developed and tested an initial prototype of a remote patient monitoring application for asthma patients in pulmonary subspecialty clinics. The original intervention consisted of a smartphone app that prompted patients to report asthma symptoms every week. The study demonstrated high patient adherence and low provider burden.

Increased access to patient reported outcomes data can aid providers in delivering patient-centered care.

For the current study, researchers adapted the intervention to the primary care setting and gathered patient and PCP feedback on requirements for a successful remote patient monitoring application.

The study’s results are based on analysis of 26 transcripts (21 patients, 5 providers) from the prior study, 21 new design sessions (15 patients, 6 providers), and survey responses from 55 PCPs.

PCP-facing requirements included a clinician-facing dashboard accessible from the EHR and an EHR inbox message preceding the visit. Nurse-facing requirements included callback requests sent as an EHR inbox message.

Patient-facing requirements included the ability to complete a one- or five-item symptom questionnaire each week, depending on asthma control. Patients also called for the option to request a callback, and the ability to enter notes. Additionally, patients suggested that the app push tips prior to a PCP visit. Requirements were consistent for English- and Spanish-speaking patients.

EHR integration of the intervention required the use of custom APIs, the authors noted.

“This study demonstrates how third-party apps can be used for PRO-based between-visit monitoring in a real-world clinical setting with the goal of maximizing use, usability, and scalability in parallel with native EHR functionality and patient portal offerings,” the study authors wrote.

“Although we focused exclusively on asthma, these findings may generalize to other chronic conditions that benefit from routine symptom monitoring using standardized PROs, such as rheumatologic disease, mental health illness, and irritable bowel disease,” they continued.

Additionally, the authors noted that their study’s findings could be applied more broadly to support primary care patient reported outcomes for patients with multiple chronic illnesses.

“Similar requirements elicitation approaches also have the potential to develop scalable interventions for monitoring overall health of patients with multiple chronic conditions, such as captured by global health PROs which measure general physical, mental, and social health,” they wrote.

“With further testing, iterative development, and continued attention to scalability, the rapidly evolving efforts of digital remote monitoring between visits may be achievable at the population level for patients with chronic conditions,” the authors continued.

As care is increasingly delivered remotely, such requirements are likely to become more important.

“Our effort is distinct from other reported efforts at developing clinically integrated remote monitoring interventions, which lack prioritization of requirements, require additional clinical staff, such as care managers, to monitor data, or require a device,” the authors wrote.

These alternate approaches may encounter scalability issues, such as cost challenges for using devices where they aren’t necessary, they explained.

“Furthermore, we provide new knowledge regarding how a third-party application can be integrated into an EHR with patient- and provider-facing components to enable the use of PROs for between-visit monitoring.”