Leslie McIntosh is Director of Clinical Informatics at Washington University School of Medicine and co-leads the St. Louis Machine Learning Group. She’ll share how modeling systems of patients, research and outcomes provides evidence to guide care decisions, and how seeing them as social actors helps make sense of really big data.
Confluence of Analytics, Information Systems/Computer Science, Domain (in this case HealthCare)Generally have:Passion for 1Training in 2Desire to learn 3rd
PeopleStarting teamHiring studentsWillingness to learn
CIDERAggregation of the electronic medical record for BJC and WashU inpatient systemWeb-based systemNot requiring specialized programming skillsTrained End-users (clinicians and clinical research managers) to use it
A Few people used the systemBut mostly not.Moving data to information is a challenge in translational researchBuilding a tool is not sufficient for non-informatics researchers to access meaningful dataMost clinical researchers think in terms of individual healthMost public health researchers are not aware of the data elements in the medical recordCreating a bridge between the tool and researchers provides a chauffeur in moving data to information
Computer science/Information ScienceAnalysts & StatisticiansClinical Researchers
Tend to focus on the software being built from the requirements presented not from the end-users needs
Technical users of the systems built by developers, Hypothesis Agnostictypically in love with dataThose people that love data rarely love people
Many ideas and depth in clinical specialtyResearchers are not generally trained in data collection, management, and analysis.
In the end we have, developers, analysts, and clinicians trying to get data to informationIn my setting…Developers tend not to focus about the meaning of the dataAnalysts don’t understand computer programmingClinicians who don’t take ownership of data
A common thread is data…Struck oil…what do we do with it now?Data are the oil of scienceIn the social network sense, the people who have data are the brokers,Like in social networks, without a broker or bridge, these are isolated nodes in a social network
Varying levels of education (from Phd, to Masters, Undergraduate, and students)
SkillsTechnicalA statistical tool (R or SAS)Scripting language (Python)MedicalMinimal understanding of medical recordResearch
Investigative nature(Sherlock Holmes, without being a sociopath)E.g., Need to determine why data are missing: query is wrong, data are not available, laws changedCommunicationWithin team (fail early, fail often)With investigators
From unstructured and structured medical documents
NSF & NIH (National Science Foundation & National Institutes for Health) funding initiatives for Big Dat and for the first time, big data analystsAcademia &