Precision medicine — an approach that uses information about a patient’s genes, environment and behavior to personalize disease prevention and treatment — relies on access to high-quality data. By analyzing large amounts of information from EHR and other sources, precision medicine has the potential to transform various healthcare and medical research applications, including organ-on-a-chip technology, stem cell therapies and cancer care.
Analytics-driven precision medicine in oncology has already generated considerable interest in the healthcare and life sciences sectors, but curating and processing the data needed to inform these efforts remains a challenge.
To address this issue, Precision Health Informatics (PHI), a subsidiary of community cancer care provider Texas Oncology, is collaborating with health technology company COTA, Inc. to pursue AI-driven data curation aimed at accelerate precision medicine at the point of care. .
The challenges of data curation in precision medicine
Part of what makes processing data difficult in the context of cancer care is the wealth of information that healthcare organizations have access to through the proliferation of EHRs, according to CK Wang, MD, chief medical officer of COTA.
He emphasized that although the concept of real world data (RWD) is not new, the accessibility of medical records data is a more recent development. Before the advent of EHR systems, much of RWD in healthcare came from claims data and prescription information, which limited the information that could be derived for use in the clinical sphere.
Wang emphasized that EHR data gives clinicians and researchers unprecedented access to a wealth of new information about patients and models of care. However, this increased availability of RWD comes with its own pitfalls.
First, the scope of what constitutes RWD has continued to expand alongside the availability of tools such as wearables and capture frames. patient-reported outcomes. Additionally, data from sources such as EHRs can be “complicated,” requiring approaches to parse this information for data analysis purposes.
“Most of this data resides in unstructured information,” Wang explained. “More and more over the years — although there are discrete pieces of data (in) the more structured data, that you could extract pretty quickly — the information that we’re looking for when we’re talking about real clinical data continue to appear.
For the foreseeable future, even with the rapid evolution of technologies like AI, there will still be a human component to this job because you need that human expert.
CK Wang, MDChief Medical Officer, COTA, Inc.
He noted that to date, extracting unstructured data from medical records to make them more usable requires considerable expertise and human resources. COTA’s work addresses this abstraction of the EHR and the company has developed an AI-driven data curation tool, known as CAILIN, to streamline this process and allow users to query a dataset like they would do it in a search engine.
But even with such tools, Wang said the role of an expert with a medical background is essential to abstracting clinical data.
“For the foreseeable future, even with the rapid evolution of technologies like AI, there will still be a human component to this work because you need that human expert,” he said, noting that these tools are intended to alleviate some of the burdens. of manual data abstraction, rather than replacing the humans involved in the process.
Wang also noted that when using data to inform analytics efforts or develop algorithms, data curation challenges give way to data quality issues. The adage “junk in, junk out” is often brought up in conversations around AI technologies to reinforce the fact that the quality of an algorithm depends on the quality of its training data.
In healthcare, this is particularly important both in and out of the AI development space, as the creation of care guidelines and treatment paradigms draws on lessons learned clinical trials.
Lori Brisbin, PHI’s chief operating officer, explained that in the context of precision medicine, including RWD alongside clinical trial data is particularly useful because the parameters of a clinical trial do not necessarily reflect the scenarios of the real world. Only patients meeting a rigid set of criteria might be eligible for a drug trial, but in practice other groups of patients might also benefit from the use of that drug, albeit at a different dose or in combination with other medications.
Using the high-quality data needed to understand such patients and provide precision care is at the heart of PHI’s partnership with COTA.
PHI has a database that represents approximately 1.6 million patient journeys from diagnosis to treatment. Brisbin noted that much of this information is structured, but stored in disparate sources. Then there is unstructured information, such as clinician notes, which contains valuable information but is difficult to analyze.
To get a complete picture of each patient’s journey, PHI works with COTA to curate structured and unstructured clinical data.
Building meaningful partnerships
Brisbin emphasized that choosing a partner to help PHI improve its data curation and processing came down to considerations of clinical sense as well as data analysis expertise.
“(COTA) had a very, very strong clinical knowledge base in medical oncology, and they were able to review records to make recommendations to us if there were gaps in the data,” she said .
Wang noted that providers are generally aware of the value of their data in helping improve operations and quality of care, but often face barriers when deciding whether to work with a partner or undertake data retention initiatives themselves.
As with any data-related project in healthcare, the investment in staff and resources can be costly for providers. But another barrier lies in the organization’s data curation capabilities, which he differentiates from data abstraction.
He explained that data abstraction can be understood as identifying and extracting data elements from the medical record, which is simpler than data curation, which prioritizes data quality considerations parallel to the extraction.
Some vendors may not have a good idea of the quality of data they want to extract, making collaboration with external partners with data quality expertise potentially valuable.
Reporting data quality issues and gaps is particularly relevant to cancer care. Wang said survival data, for example, is important for assessing oncology outcomes, but if a health system is missing a significant portion of that data within its patient population because deaths were not reported , this data may be unusable to effectively measure results.
However, if the organization works with a data partner, it may be able to fill gaps in its EHR data using third-party sources. Advances in interoperability and EHR ecosystems could also help this process, but the national scale of these challenges makes them unlikely to be resolved quickly.
“The foundation of this partnership is data and data analytics — the potential value of the data to give PHI and Texas Oncology insight into their patient population — and for them to then stack that on their computer technology to meet their needs. needs,” Wang said.
Bringing AI to precision medicine
The partnership helps PHI conduct hypothesis-driven studies and enrich clinical trials, which are critical to advancing cancer care.
Brisbin noted that PHI’s attrition rate — a metric that quantifies the loss of participants in a study — has improved significantly since partnering with COTA to use AI-driven data curation.
“We have the lowest attrition rate with COTA compared to any other data aggregator. So that means if we send more than 100 records, they use 87 of those records – with no missing data elements – (for contribute) to a study “It’s extremely high,” she said, noting that some other potential partners initially considered by PHI had attrition rates of around 50%.
AI also allows PHI to streamline clinical trial inclusion. The inclusion and exclusion criteria for these trials are rigorous: factors such as comorbidities are just one of many that could prevent a patient from participating in a trial.
Identifying patient eligibility for a trial requires going through entire medical records, but with well-organized data, this process is simpler. Adding AI capabilities further streamlines these efforts by enabling keyword search and other workflow enhancements that make determining eligibility less burdensome.
Brisbin emphasized that AI serves to reduce the burden of administrative tasks on PHI staff.
“If you apply AI to search for photos of dogs, you’ll get all kinds of dogs. So, let’s say you have to limit yourself to just photos of German Shepherds,” she said. “You’re going to have German shepherds, but you’re probably also going to have wolves or huskies or even coyotes. You’re going to have all kinds of things that will require someone to look at them and say, ‘No, that’s a coyote , not a German Shepherd.'”
“That’s what we’re saying: AI is going to refine (clinical data) for us and make people’s jobs easier, and just allow experts to focus on that higher level of experience and allow them to work at the highest level,” he added. ” she concluded.
Shania Kennedy has been covering news related to healthcare IT and analytics since 2022.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.Ok
