Recognised as one of America’s best pediatric facilities by U.S. News & World Report, the Children’s Hospital of Orange County is becoming a world-class destination for children’s health.

Since 1964, CHOC Children’s in Southern California has nurtured, advanced and protected the health and well-being of children through innovative care and state-of-the-art facilities.

What started as a 62-bed children’s hospital – the first in Orange County – has grown into a pediatric health care system with two hospitals and several primary and specialty care centers serving multiple counties. As part of the commitment to provide the highest quality of care, President and CEO, Kimberly Chavalas Cripe is positioning CHOC to become a leader in transforming pediatric healthcare through the use of artificial intelligence. By utilizing all available data to inform clinical decision-making, a talented team at CHOC is working to develop intelligent data science models for use by its providers. Chief Health Information Officer, William Feaster, MD, MBA and his team are moving from a model that simply reports data to providers, to application of sophisticated machine learning algorithms covering a variety of patient care scenarios.

In partnership with CHOC’s Chief Intelligence and Innovation Officer, Dr Anthony Chang, MBA, MPH, MS, Dr Feaster will continue to add data scientists to the team, further accelerating intelligence projects. Combining the vision and leadership of CHOC’s Sharon Disney Lund Intelligence and Innovation Institute with the resources and expertise of its Information Technology Department, CHOC is creating a dynamic environment to drive predictive analytical models for application in care settings.

Both Dr Feaster and Dr Chang are seeking to rapidly remove limits to analyzing and applying data to medicine. A primary goal is to provide timely access to medical intelligence and the CHOC team is working to continually decrease turnaround times required to develop and test algorithms. “What previously took three to six months was decreased to three to six weeks,” says Dr Feaster. “And soon it will be as short as 3-6 days, allowing for a significantly higher number of applications.”

There are several notable projects that the CHOC team is tackling. Led by CHOC data scientist, Louis Ehwerhemuepha, PhD, the team has developed a predictive model to designate low, moderate and high risk of 30-day hospital readmission. These predictors allow the care team to more effectively focus resources on patients with the highest risk of readmissions. The adoption of the model, and associated interventions on patients at risk of readmissions, has resulted in recent drops in hospital readmission rates. The team has also developed a model for predicting severe sepsis among patients presenting to the emergency department. This model has been accepted by the corresponding clinical team, and work on implementation and deployment has begun. It is expected that in addition to early detection of patients at risk of severe sepsis, the model will help improve the quality of care of patients requiring hospitalization. Under the research of CHOC Neurologist, Sharief Taraman, MD, models for predicting post-concussion syndrome are also in development. Once completed, the model will further improve understanding of why some patients with concussions go on to develop post-concussion syndrome and predict patients most at risk of such deterioration.

CHOC is also currently developing several artificial models with the aim of improving the quality of care ofpatients. This includes studies on:

1. Application of artificial intelligence in the treatment of preterm infants with patent ductus arteriosus
2. Automated discovery of lesions in medical images using the latest in unsupervised anomaly detection with Generative Adversarial Networks
3. Application of deep neural networks

(including recurrent neural networks, RNN, gated recurrent units, GRU, and long short-term memory networks, LSTMs) in natural language processing for sentiment analysis of clinical notes as well as patient experience free-text feedback The implementation of Cerner’s HealtheDataLab, which solved limitations in storing increasingly large amounts of data, ensures ease of access for computational applications through a secure cloud-based setting. “This machine learning environment will allow our team to dramatically expand variables and compare their relationships to levels that were previously nearly impossible to achieve”, says Feaster. CHOC’s ultimate objective is to share their analytics models with other children’s hospitals for wider application.

Additional projects include analysis on the use of opioids, predictors of patient experience, prediction of exacerbation of asthma, analysis of population health initiatives and impact on the quality of care among patients with asthma. The team also supports several doctoral students from the Computational and Data Science program at Chapman University as well as a doctoral student in nursing from UCLA. Several collaborations include professionals from UCI and Rochester University.

Dr Chang summises, “As we continue to remove barriers, shortening time frames for development of datarich algorithms, and move toward prescriptive implementation of intelligent data models to transform pediatric healthcare, the sky is no longer the limit.” With innovation being one of its core organizational values, CHOC established The Sharon Disney Lund Medical Intelligence and Innovation Institute (MI3), a unique and first-of its-kind institute that creates, focuses and executes projects in the areas of intelligence and innovation in pediatric medicine. These two interrelated disciplines, extremely limited in development in the pediatric realm, hold great promise to change the trajectory of pediatric care around the world. MI3 aims to foster robust developments in artificial intelligence methodologies, as well as innovative advances in emerging areas such as genomic medicine, regenerative medicine, robotics, nanotechnology
and medical applications/devices.

MI3 is dedicated to empowering data intelligence and medical innovation at CHOC, as well as driving innovation leadership in the international pediatric community. Meanwhile, CHOC continues to make waves with its MI3 Internship Program – a unique program for high school and college students now in its seventh year. Such is its reputation, this year over 350 applications were received with 63 being selected by Dr Sharief Taraman, the Director of the program, and his team. The interns shadow physicians, are mentored by data scientists and a large group of CHOC providers and associates. The program provides invaluable experience and a potentially life changing opportunity for high school students. One of whom has already become a medical innovator in her own right.