Development of a secure, cloud-based platform to improve record linkage & cross-agency collaboration for the public sector: using deep learning & scalable data integrations to combat the opioid crisis

Project Summary AbstractThis SBIR Phase I proposal aims to fund research and development for a newmultitenant secure cloud based platform specifically tailored to provide local governmental agencies with tools to share datasets and link them accuratelyat high quality and low costThe OpenLattice platform will focus on reducing drug overdoses and making drug treatment less fracturedIndividual level datasets linked across medical providers and law enforcement can support analyses of prescribing pathways and treatment trajectories that precede opioid overdoseentry into treatmentdisruptionand recoveryHoweverlinking data at the individual level has proven to be a difficult and resource intensive endeavor compared to use of aggregate level datawith issues with deduplication plaguing many institutional databasesWithAmerican deaths recorded daily from opioid overdoses and systems of care spread across multiple institutionsthe need for greater and high quality data sharing is undeniableOur test partner for assessing the efficacy of proposed innovations is the Greater Portland Addiction CollaborativeGPACin Mainea partnership of hospitalsa police departmentjaildetox treatment centers and halfway houses already working together to reduce drug overdosesThis proposal aims to demonstrate proof of concept foriscaling high quality data integrations across multiple governmental domains via a standardized entity data modelandiiimproving record linkage using neural networksFirstlyOpenLattice is developing an open source ontology and integration scripts to standardize integration of datasets into OpenLatticeandapos s databaseAs the individual customization requirements decline for onboarding customers and integrating new data into the platformcosts will be greatly slashedremoving a significant barrier to data solutions for smaller counties and cities across the countrywho have historically faced custom integrationssystem updatesdata storage fees and add ons at high costThe OpenLattice platform also enables use of existing ETL tools and seamless integration with police dispatch systemsemergency medical callshealthcare recordsand online prescription systems across partners who have committed to data sharing and collaborationSecondlyOpenLattice is developing a newproprietary algorithm for record linkage that employs a promising but as yet commercially untested techniquea multilayer perceptron neural networkmore commonly known as deep learningIn pilot researchthe linking algorithm has already demonstrated success rivaling and sometimes exceeding current state of the art linking technologiesIn Phase IOpenLattice will continue to improve ontologiesintegration toolsand the deep learning neural networkand test on publicly available datasets with dissimilar data types and formatswith manual confirmation of resultsWhen successfulthese innovations will address critical barriers to improving clinical practice in treating opioid addiction by enabling a more comprehensive continuum of care for those in treatment Project NarrativeLarge scale and coordinated responses to several of the US s hot button public health and criminal justice issuessuch as the opioid epidemic and mass incarcerationare complicated by poor resource sharing and the US government s highly fractured jurisdictional authorityThis Small Business Innovation Research Phase I project aims to develop an efficientscalablecloudbased platform for hosting and linking highly sensitive state and local government databases at low costusingiinnovative data integration scripts and ontologies that standardize and scale capacity andiitechnical advances in record de duplication for linking databasesData solutions would have tremendous societal impact on understandings of public health and the opioid epidemic by making drug treatment less fracturedsaving lives and dramatically broadening contextual informationonce data is broken out of silos