Clinical OMICS

SEP-OCT 2018

Healthcare magazine for research scientists, labs, pathologists, hospitals, cancer centers, physicians and biopharma companies providing news articles, expert interviews and videos about molecular diagnostics in precision medicine

Issue link: https://clinicalomics.epubxp.com/i/1023557

Contents of this Issue

Navigation

Page 29 of 51

28 Clinical OMICs September/October 2018 www.clinicalomics.com tations that allow us to find patients within the EHR that share certain characteristics," said Freimuth. "We can do this to help identify patients for clinical trials." Krevsky Elkin said N-of-One has done a lot of work to develop decision support tools that curate a list of trials and filter them based on the clinical and molecular criteria, geographic location, and strength of evidence surrounding the therapy. But EHR data are not currently incorporated. "At this point, we are casting a wide net in terms of the clinical character- istics because we don't pull in those specific patient data from the elec- tronic medical record. But our hope is that, sometime in the near future, we'll be pulling in that data and able to filter the trials based on that infor- mation as well." Emerging Role of Artificial Intelligence The U.S. Food and Drug Administra- tion approved the first AI-based CDS system from artificial intelligence company Viz.ai earlier this year for stroke. Although AI may have a place in CDS systems one day, it's still an emerging field focused more on using AI to aid, rather than super- sede, physicians. "The approach we're taking to AI is really about orchestration of clin- ical care," said Culot. For example, Philips offers Illumeo, a data analytics tool that uses adaptive intelligence, a subset of AI. Illumeo understands the anatomical context and identifies the appropriate tools the radiologist needs to make a diagnosis, similar to a surgical technician handing a surgeon the right instrument at the right time. "It's a case of using AI as a clinical decision support tool, but (continued from previous page) Novel Algorithm Identifies Epilepsy Surgical Candidates Sooner Epilepsy is one of the leading neurolog- ical disorders in the United States. Most patients with epilepsy can be treated with medication, but for about 30% of patients, medication does not provide sufficient control and surgery is the next step. At Cincinnati Children's Hospital, research- ers developed and implemented a novel natural language processing (NLP) algo- rithm that examines the linguistic chang- es in clinical progress notes over time to identify pediatric patients with epilepsy who may be eligible for a surgical con- sult. NLP, a form of artificial intelligence, is a common way to extract meaningful, actionable insights from the wealth of unstructured data in the clinical notes of electronic health records (EHRs). Thanks to a two-year, $250,000 grant, researchers implemented the algorithm at Cincinnati Children's in 2016, and have since shortened the time from diagnosis to surgery by about two years for their pe- diatric patients. At Cincinnati Children's, the average time from diagnosis to sur- gical intervention is six years, and the na- tional average is ten. Lead investigator on the project, Judith Dexheimer, Ph.D., assistant professor in the division of emergency medicine at Cincinnati Children's, said, "Everything runs automatically right now, so there's no additional human intervention." The algorithm is configured to auto- matically examine the clinical progress notes in the EHR on Sunday nights and send providers an email and an in-basket message in the EHR that says a particu- lar patient may be eligible for a surgical consult. After alerts are sent, the research team follows up with providers about whether they placed an order for a sur- gical consult. Frequently, for those that didn't place an order, there were other comorbid conditions that could not have been predicted based on EHR data. "The eventual goal will be to evaluate the classifier, and move this same sys- tem into the adult side [at the University at Cincinnati]," said Dexheimer. Although there are no specific plans yet, she added the hope would be to collaborate with and implement the system at other insti- tutions. —Christina Bennett n Judith Dexheimer, Ph.D., assistant professor, division of emergency medicine, Cincinnati Children's Hospital Thanakorn Phanthura / EyeEm / Getty Images

Articles in this issue

Links on this page

Archives of this issue

view archives of Clinical OMICS - SEP-OCT 2018