n psychiatry, clinical judgment has been the predominant method of diagnosis. The `gold standard” for
clinical research is the Structured Clinical Interview for DSM-5 (SCID-5). The SCID-5 promotes reliability, but
its validity is questionable. There is growing recognition that biomarkers can be used to identify more
homogeneous patient populations, but there is substantial phenotypic discordance for schizophrenia even in
identical twins. In the absence of symptom data, biomarkers alone are unlikely to yield a clinical diagnosis.
In a past SBIR project, TeleSage Inc. successfully converted the paper SCID into a web-based
software program. The NetSCID-5 is now widely used in research. Nevertheless, in order to achieve the goal
of “turning clinical care networks into centers for research” (Insel, Director's Blog, 2012) and furthering the
RDoC initiative, we need a means of gathering rigorous diagnostic data in routine clinical care. This software
must (a) require minimal clinician time, (b) allow clinics to bill according to the current DSM-5 categories, and
(c) be able to gather the large amount of data necessary to inform a broad research agenda including machine
learning techniques. TeleSage proposes to develop a self-report diagnostic assessment that satisfies both
immediate clinical needs and broader research goals: the “Screening Interview for Diagnosis” or SID.
TeleSage has worked with an expert panel including Dr. Michael First, the primary author of the SCID-
5, iteratively developing and testing self-report items. Based on expert panel review and cognitive interviewing,
we identified a final set of 661 unique self-report, Likert-scale items covering all of the individual sub-symptoms
described within each of the SCID criteria. TeleSage has also created a behavioral health PORTAL that
includes the NetSCID-5, IRT/CAT item administration, randomization, and longitudinal reporting capabilities.
The PORTAL exchanges data and reports with several EHRs including the NetSmart EHR system.
This Direct-to-Phase II application aims to create the SID, which will (a) reside on our existing secure
web PORTAL, (b) administer simple Likert-scale self-report items, (c) generate DSM-5 and ICD-10 diagnoses
for billing, (d) use minimal clinician time, (e) pull pre-defined data fields from the EHR (e.g. family history,
demographic, and biomarker data), (f) be able to add new self-report items for exploration, (g) integrate with
machine learning tools, and (h) send raw data and interpretive reports to EHR systems. Because of these
features, we believe that the SID has the potential to be used widely by both clinicians and researchers.
The SID is intended to help transcend DSM-5 and to support RDoC. The SID product must be
commercially successful and useful both in routine clinical care and research. The SID is intended to facilitate
the development and evolution of a new behavioral health nosology based on the aggregation of biomarker
and symptom data, a nosology that is more analogous to those found in other fields of medicine.