Using data analysis to improve research, patient care and clinical outcomes
A tremendous amount of electronic data is available to improve children's health care and influence lifetime health. The Johns Hopkins All Children’s Health Sciences Research Informatics brings innovation in using electronic health data to provide high quality care—from achieving the best outcomes to ensuring the highest level of patient safety and lowering the cost of care.
Health Sciences Research Informatics provides the tools and knowledge to use data more effectively institution-wide in real-time clinical decision-making, pediatric research and other operational decisions that can enhance patient care. Experts in research methods, database design and integration, epidemiology, biostatistics and predictive analytics provide meaningful data and analysis that improve care and increase understanding of pediatric illnesses.
Multidisciplinary collaboration is integral to the efforts of Health Sciences Research Informatics. Our team works closely with researchers and clinicians to use data-driven information to better understand the underlying causes and mechanisms of diseases, as well as to predict or determine effective treatments in specific patient populations—the essence of precision medicine.
The work of the Health Sciences Research Informatics team enhances patient care in real-time—from something as simple as creating an alert to remind the medical team to redose a child with antibiotics during a 12-hour surgery or as complex as shaping guidelines that eliminate costly blood typing and matching for patients who don’t need it.
The Health Sciences Research Informatics team also works closely with Decision Support (Finance Division) and Information Technology leadership to continually develop the enterprise Data Warehouse and initiatives that enhance value-based care.
Epidemiology and Biostatistics Unit
The epidemiology and biostatistics teams within Johns Hopkins All Children's Health Sciences Research Informatics provides collaboration and consultation services on study design and data analysis to the hospital's investigators and their collaborators.
We assist investigators at each stage of a study’s lifecycle, including refining the research question, study design, mid-study evaluation and the analysis, interpretation and reporting of results. Services are matched to the needs of each investigator and duration, from long-term collaboration to routine service support to enhance the research efforts of investigators.
Biostatistics services are also available for quality improvement and operational needs and projects. The biostatistics team provides analytic services for all areas across the organization.
Database Design and Data Management Unit
The Database Design and Data Management team provides collaboration and consultation services on database design and data management to assist investigators and stakeholders at Johns Hopkins All Children’s Hospital and their collaborators.
The primary objective of the Database Design and Data Management team is to assist in the design of data collection tools, design and build of database architecture, query and report design and execution, and ongoing data management assistance throughout the life of a research study, operational endeavor or quality improvement initiative.
The Database Design and Data Management team provides support for the use of electronic survey tools such as Qualtrics and also serves as the administrators for the use of REDCap at Johns Hopkins All Children's.
Specific types of support include, but are not limited to:
- Database design: Creation of the database and corresponding web-based case report forms (CRFs) according to the Institutional Review Board (IRB)-approved or other protocol specifications regarding confidentiality and protected health information, with appropriate enforcement of data security through the implementation of user-based accounts and permissions.
- Survey design: Electronic platforms for survey distribution and deployment, as well as assistance with survey design to reduce response bias and ensure valid results.
- Database management: Electronic storage, governance, and oversight for databases that are required for research or quality improvement studies.
- REDCap design and implementation: The primary electronic data capture platform supported by the team is REDCap, a secure web application designed by Vanderbilt University specifically for data capture of research studies and operational or quality improvement endeavors. Qualtrics is also supported for direct electronic surveying, where needed.
- User support and training: User training at the outset of the study and support on a continuing basis, as needed.
Clinical Data Acquisition Unit
A team of Analytic Specialists within the Institute for Clinical and Translational Research provide expertise on sourcing, extracting and formatting clinical data from the Johns Hopkins All Children's Data Warehouse and other clinically-focused data repositories. The team gathers, consolidates, and analyzes data for use in clinically-oriented research, quality improvement projects and operational processes.
Information housed in the Data Warehouse also supports Johns Hopkins All Children’s population health initiatives to improve the health of all children, through the analysis of data related to the influence of genes, environment and treatment options for common child health problems like prematurity, asthma and obesity.
The Data Warehouse team adheres to Johns Hopkins Data Trust standards and serves as the adjunct Core Clinical Data Acquisition (CCDA) and honest broker, maintaining strict quality control as well as privacy and security safeguards to ensure the validity of all member-related data.
Predictive Analytics
The predictive analytics team is focused on advancing the use of artificial intelligence and machine learning at Johns Hopkins All Children’s. We collaborate with all groups across the hospital, with the overall goal of improving quality and safety, increasing patient satisfaction and reducing cost.
We also develop visual analytic solutions by leveraging data from a wide variety of medical, biomedical, and public health repositories. Our team includes experts in machine learning modeling, computational physiology, hyperparameter optimization, model deployment, model testing and validation against blind datasets.
We work with all areas of the hospital, using data to discover solutions to complex problems.
The predictive analytics team at Johns Hopkins All Children’s Hospital helps to find optimal solutions to complex problems by analyzing the immense amount of information compiled every day in 21st century health care and hospital records.
Predictive analytics incorporates a variety of quantitative methods and techniques—including statistics, mathematics and computer science along with technology (both hardware and software)—to develop algorithms designed to discover new patterns and make predictions about what may happen in the future, based on the trends we can observe in current or historical data.
The predictive analytics team at Johns Hopkins All Children’s
We deliver predictive analytics solutions in domains that include clinical decision support, chronic disease management, readmission prevention and adverse event avoidance.
Luis Ahumada, Ph.D., predictive analytics director, with Ali Jalali, Ph.D., predictive analytics data scientist, leads a team of doctorate-level data scientists. The team has a wealth of experience in managing large datasets, working with advanced data analysis frameworks and formulating novel data management and modeling methodologies.
Improving the future of health care now
Predictive analytics uses advanced mathematical methods, powerful statistical models and machine learning algorithms to represent, interrogate, and interpret data that is frequently too complex for human intelligence alone. There is great interest in the health care field in using predictive analytics techniques and artificial intelligence methods such as machine learning and natural language processing to teach computers to learn for themselves how to discover and capture hidden relationships in data.
These data come from many sources and may include the electronic health records, intelligent medical devices, laboratory tests and monitoring devices. Using all the elements of predictive analytics, we can forecast trends, predict patterns or unknown outcomes, produce estimations of likelihoods or parameters, generate classification labels, and contribute many other aggregate or individualized forecasts.
Often the unknown event of interest is in the future, but we can apply predictive analytics to any type of unknown whether it be in the past, present or future. The core function of predictive analytics relies on discovering unknown relationships between explanatory variables and predicted variables by analyzing past events to predict the unknown outcome. The accuracy and usability of results depend greatly on the level of data analysis and the quality of the input data. In health care, predictive models exploit patterns found in historical and transactional data to identify risks and opportunities.
Predictive Analytics at Johns Hopkins All Children's Hospital