This Funding Opportunity Announcement (FOA) invites researchers to submit applications for multi-disciplinary diagnostic strategies for rare diseases that combine machine-assistance, genomic analysis, and clinical consultation. Of particular importance, these strategies must be able to be adopted and performed at the primary or secondary care levels by front-line healthcare providers and be readily integrated into their clinical care workflow.
There are an estimated 7,000 different rare diseases, most of which (~85%) are monogenic disorders, many of which are very low prevalence disorders (<3,500, or fewer, patients in the US), and most have considerable within-disease phenotypic heterogeneity. Given the low prevalence, most front-line clinicians may have no prior experience with the individual diseases, which contributes to the difficulty in diagnosis, and often requires specialist, sub-specialist or multi-disciplinary referral to accurately diagnose the patient.
Current diagnostic approaches for hard-to-diagnose patients, many of which have rare diseases, are typically made through idiosyncratic specialist and sub-specialist referrals, often located at tertiary care and/or academic institutions. Specialists and sub-specialists are a scarce resource that may result in substantial time delays in obtaining appointments, and for some patients, may require travel of long distances for evaluation. Because rare disease and hard-to-diagnose patients often require multiple specialist referrals, and typically multiple tests and procedures, there can be substantial time delays in obtaining a diagnosis.
Genomic analyses are not routinely performed in clinical practice. In recent years, genomic testing has become more available, and for more commonly encountered conditions, a diagnosis may be readily made in the appropriate context. However, interpretation of genomic analyses for rare monogenic diseases is difficult, and often requires subspecialty evaluation and genetic counseling referral. These are also scare resources contributing to diagnostic delays.
Additionally, knowledge accumulation in rare genetic diseases is rapidly advancing; clinicians need access to accurate and up-to-date information that is readily available, and able to be integrated into clinical practice to facilitate rare disease recognition. The volume of this information is nearly impossible for clinicians to manage and machine-assistance to facilitate decision support has been recommended as an area of interest and development; however, currently most decision support is focused on common disease approaches, such as cardiovascular disease algorithms, and safety warnings for drug prescribing and interactions. Rare disease tools are being developed, but most are stand-alone programs that require clinician awareness, often require substantial data entry or training to use, and are not integrated into usual care. Similarly, voluminous information is available in healthcare system databases and electronic medical records (EMR), but the management of this information to facilitate diagnosis is challenging.
Thus, developing better diagnostic strategies that could incorporate clinical, machine-assisted and genomic analyses that could be readily integrated into front-line clinical care is likely to provide more rapid identification, escalation, and accurate diagnosis of hard-to-diagnose patients.
Clinical approaches for these projects may include multi-disciplinary expert diagnostic teams, comprised of, for example, clinical specialists, informatics experts, geneticists, and other subject matter experts, who would work collaboratively to diagnose patients referred by front-line physicians, or the development of frameworks through which front-line providers can escalate a suspected rare disease patient to a multi-disciplinary diagnostic team.
Artificial intelligence, including machine learning or other information technology (IT) (collectively referred to as “machine-assistance”), components of the project may include the development of algorithms that computationally make predictions based on data. Machine-assistance strategies may be applied to the EMR or other healthcare system databases, genomic data, imaging data, and other biological domains. Methods may include knowledge extraction, such as natural language processing; or machine capture and interpretation, such as facial recognition. The goal would be to develop and apply algorithms that could identify potential rare disease patients on the basis of, for example: medical utilization patterns (e.g., high-utilizers, young age); sentinel characteristics or other features (e.g., abnormal gait, facial features, delayed development); imprecise diagnosis (e.g., neurologic disorder not otherwise specified, failure to thrive); or based on clusters of diseases that are related in some way (e.g., generalized seizures, motor impairment).
URL for more information:
Filed Under: Funding Opportunities