Grant proposal program: enhancing funding attraction for clinical research


The SporeData Grant Proposal Program designed to assist faculty members with funding attraction for clinical research. Briefly, we will locate funding opportunities, propose ideas for specific projects, and work iteratively with you while we write the full Methods section. SporeData is added as a subcontractor for the services provided through the funded project, and all costs are discussed upfront. We provide further details in the following sections:

steps

Workflow

  • We will send information on upcoming funding opportunities that match your specific clinical research interests along with ideas for projects making use of the latest Data Science technologies. The objective of these proposals is to generate clinically-relevant information. Funding sources might include:

    • Federal (NIH, PCORI, DoD, among others)
    • Companies (imaging, device, pharma)
    • Foundations and medical societies
  • If you like the initial idea, we will write a Specific Aims section and also send you a cost ballpark.
  • If you agree with the Specific Aims, we will then proceed to write a full Methods section. The faculty member will then add a Background Section (which can often be recycled from previous proposals), and submit the completed proposal as the PI.
  • Once the first proposal is completed, we will then identify other funding opportunities that might be interested in different proposals following the same overall design.

A typical faculty member is often able to increase his/her number of annual submissions by approximately ten new proposals. This program significantly increases the faculty member’s ability to attract funding as well as their output of high impact publications.

Opportunities for multi-institutional collaboration

Proposals involving more than one institution are often more appealing to funding organizations, as multi-institutional research teams tend to aggregate more resources and critical mass. As a result, SporeData attempts to link clinical researchers with similar interests.

The workflow for this connection is identical to the one outlined above, except that during the initial discussion regarding the project objectives we will also bring up the possibility of adding some of your existing collaborators from other institutions. Alternatively, we might suggest individuals outside of your current professional network, based on their expertise and potential contribution to the project.

Data science methods

SporeData offers a complete set of traditional statistical methods in addition to a wide range of sophisticated Data Science methods. A select sample of these methods is displayed below:

  • Machine learning: prediction of clinical outcomes and costs for individual patients, also allowing for the creation of Web applications for use during the interaction with patients.

scatterplot

  • Geographical Information Systems: mapping conditions, procedures, and outcomes for a given geographic area.
  • Personalized outcomes scales: computerized adaptive testing, similar to PROMIS or SAT, but measuring a wide range of constructs and also available through Redcap. Main advantages include a smaller number of questions per assessment, greater precision (reflected in smaller sample sizes), and the ability to compare scores from different scales.
  • Bayesian Adaptive Trials: similar to traditional randomized trials, but requiring a sample size that is, on average, 30% smaller, and allowing for multiple interim analyses to keep you updated on its results on a regular basis.
  • Natural Language Processing: extraction of information from free text (radiology and pathology reports, surgical and discharge notes) to be analyzedusing regular statistical methods.
  • Cost variation analysis and evaluation of low-value services: using a variety of visualization methods combined with machine learning to identify sources of healthcare services that do might not add value to patient care, also offering an opportunity for cost optimization.
  • Deep learning for imaging recognition: applying methods to read medical images (MRI, CT, radiographs) directly through software that can detect patterns for diagnosis or outcomes prediction. Compared to traditional exam reading by humans, Deep Learning can extract a wider range of findings as well as achieving greater reliability (no variation across observers).

pirateplot

Contact us at contact@sporedata.com