Opportunity Alert – AC-150513-054 – RPM/Rural Health SBIR Grant
Date Sourced: March 28, 2016
Category: Funding Opportunity
Status: Active Discussion
Potential (High-Moderate-Low): High
Add to SFA (Y/N): TBD
Client Action Required (Y/N): No
Opportunity:
University of Texas Nursing School is interested in collaborating on a
rural health diabetes remote patient monitoring research grant application.
Summary of Activities/Tasks:
1. March 28, 2016 – Coordinating discussion with Jing Wang, University of
Texas Health Science Center at Houston (UTHealth), Associate Professor,
Department of Nursing Systems, School of Nursing; and Adjunct Associate
Professor, School of Biomedical Informatics.
2. April 5, 2018 – Conference call with Jing Wang to discuss the research
aims, funding vehicles, and application premise. Call went about 90 minutes
and we concluded that an SBIR application provided the best funding
opportunities and limited the complexity of getting UTHealth support.
Given the maturity of the RPM platform, we are looking to apply for a
direct to phase 2 application which increases the funding range to $1M
(versus $150K for phase 1). The extended effort beyond the phase 1-2 is an
R grant with funding potential between 3-10M depending on the
innovativeness of the application. More discussion to follow on the topic
of “innovation” and “significance.”
3. April 19. 2016 – Outlined the core aims and basic research plan for
discussion with Jing Wang. Awaiting response and coordination of follow-up
call to refine the research scope prior to reaching out to NIH for
validation.
Forward Statements and Guidance:
Ron is currently drafting the preliminary research aims to be reviewed and
further refined by Jing. Draft is to be completed by 4/22/16. SBIR
application due date is 9/5/2016.
End.
Return to Listing
COMMENT FORM.
Comments
Facilitator
Summary of the potential aims for the study:
“Prove that real-time biometric data can be used to personalize health and wellness care plans and/or initiate preventative care interventions – with the goal of achieving demonstrable improvements in patient overall health (e.g., reduced comorbidity, comparatively less ER visits, improved health metrics [a1c, weight, etc.], slowed disease progression compared to control group, etc.).”
Another suggestion is to focus on “personalization.” NIH has a strong interest in using data to personalize healthcare, yet effective real-world applications are hard to come by. We would develop a Cmap similar to what was done for CHF that includes diabetes risk factors and evidence-based care interventions supported by:
> the individual patients EHR as a personal contextual baseline,
> ongoing personal biometric data measures, and
> periodic review of ICD-10 codes to identify new diagnosis/findings.
This approach should give us enough to support personalized health and wellness care plans and/or preventative care interventions.
Facilitator
JIng: I like the drafted aims, and it is perfect to use personalized care to dictate our purpose, I think it was what we had in mind, just not spelled out clearly in the beginning.
Jing: I am still educating myself on the cmaps, and I will get back to you once I figure out the “HOW” part. I think the key is how can we use real-time biometric data to personalize care plans, and that’s the part we need to highlight in the aims section.
Ron: Agreed. The HOW part is a little tricky given that the aims are not based on a defined clinical protocol or standard of care, but rather more subjective quality of life type measures. It might be helpful to setup a call with our clinical analytics team to talk through potential approaches.
Jing: For ongoing review of ICD-10 codes, what’s the purpose? Finding out what complications they developed as a way to measure disease progression?
Ron: Yeah, that was my thinking. Also, if the patient undergoes a treatment that impacts their readings in some way – we could misinterpret the data due to lack of context. Decreased activity could be explained if they had leg surgery, increase in temperature if they had an infection, etc.
Jing:
Another concern I have is that patients may not go to the same hospital or clinic for different condition or treatment. So maybe embedding a function in the app to ask their recent clinical or medical visits and ask for permission to extract data would be necessary.
Facilitator
Diabetes Remote Monitoring Aims Discussion:
The basic idea is to use the select biometric device data coming from participating diabetes patients to anticipate health-related complications, predict risk, and support evidence-based interventions aimed at helping patients adopt healthy behaviors, lifestyles and health outcomes.
The premise is that collecting patient biometric data (devices) and patient reported symptoms (via surveys) alone will not result in improved patient health outcomes. Rather, there must be effective analytics supported by clinician-guided reasoning to drive appropriate clinical and/or behaviors interventions. The high level overview is as follows:
> We would first establish: (a) the types of patients to be included in the study (i.e., eligibility and exclusion criteria); (b) the core research questions to be answered during the study; and (c) the types of clinical decisions and/or interventions to be supported by the study (and by whom – pcp, nurses, caregivers, etc.).
> Develop the clinician-guided oncology: have you and up to 4 other qualified research partners outlined the basic Cmap to be used to guide the machine learning model? The Cmap highlights the relationships between patient health, biometric data and reported symptoms.
> Determine the monitoring devices: identify the monitoring devices to be provisioned and kitted for this study. Select devices must support the collection of data inputs required to run the patient analysis model. At a minimum, model inputs will include biometric device readings and patient surveys.
> Incorporate patient medical record data (optional): If approved by research partners for purposes of this study, it may be possible to include participating patient medical record and/or claims data (i.e., lab results, ICD-10 codes, prescriptions, etc.) to enhance the specificity and sensitivity of our predictive analysis.
The Cmap construct is a “tripe” which involves 2 concepts and a relationship. Development of a Cmap involves determining: (1) what are some predisposing risk factors for diabetes; (2) what are some of the manifestations that suggest diabetes is developing, or are already present and evolving; and (3) what are some of the markers of severity, increased comorbidities or complications once diabetes has developed.
The approach will need to include a range of evidence-based risk factors, enhanced machine learning algorithms, and differential diagnosis-type semantic reasoning for enhanced specificity. In other words, a model that tells a provider, patient or caregiver something they already know isn’t helpful – so the model should suggest interventions or behavioral modifications that anticipate “developing” health risks AND guide patients toward the adoption of healthier behaviors, lifestyle choices and health outcomes.
Facilitator
Summary of the potential aims for the study:
“Prove that real-time biometric data can be used to personalize health and wellness care plans and/or initiate preventative care interventions – with the goal of achieving demonstrable improvements in patient overall health (e.g., reduced comorbidity, comparatively less ER visits, improved health metrics [a1c, weight, etc.], slowed disease progression compared to control group, etc.).”
Another suggestion is to focus on “personalization.” NIH has a strong interest in using data to personalize healthcare, yet effective real-world applications are hard to come by. We would develop a Cmap similar to what was done for CHF that includes diabetes risk factors and evidence-based care interventions supported by:
> the individual patients EHR as a personal contextual baseline,
> ongoing personal biometric data measures, and
> periodic review of ICD-10 codes to identify new diagnosis/findings.
This approach should give us enough to support personalized health and wellness care plans and/or preventative care interventions.
Facilitator
Received comments and feedback from Jing. We need to clearly define meaningful aims; and present commercially viable methods and systems for the proposed direct to phase 2 application. Ron will draft a more detailed solution architecture for discussion. Looking to coordinate a Webex for Wednesday, April 27 to review.
Facilitator
Wang, Jing
Apr 21
Hello Ron,
Thanks for starting the proposal.
We have no problem getting IRB and partnering hospitals in Texas Medical Center, there are ample choices here. In fact, multiple IRB may be needed when we recruit from another hospital, but we do that all the time. We need to decide on our approach first, before we approach partnering hospital systems, because the population may vary, we have a rich patient population with very rich individuals and very poor individuals. We need to decide whether our focus is those who owns a smartphone or not.
I would avoid the wording of registry though, because 100 patient is far from calling itself a registry. I will just limit it to a randomized clinical trial testing efficacy of using an innovative analytic approach in risk prediction utilizing RPM data.
My general comments for the aims are that it is too abstract as a phase 2 proposal, reviewers are looking for specific commercialization products and what the product can do and can it improve health outcomes and save healthcare costs.
My specific comments for the aims below:
> Symptom Documentation – Document what symptoms patients are exhibition. NOT sure how RPM is capturing symptom data.
> Personalized Medicine – Predict what symptoms are likely to be experienced by a patient. MAYBE predict hyperglycemic or hypoglycemia events.
> Resource Utilization Management – Avoid acute visits by daily tracking of patient bioinformatics. AGAIN avoid ER visits due to hyper or hypoglycemia events or hypertension caused fainting
> Cost Avoidance – Reduce hospital readmission rates among high acuity patients due to ineffective monitoring of key patient levels. WE can measure hospital readmission or ER visits pre and post
> Pharmacy Surveillance – Automatically submit refills and identify necessary prescription changes. THIS needs some high level data connection, not sure whether we can find a partner.
> Economic Evaluation – Assistance in quantifying the economics of diabetes treatment by taking into account management of comorbidities and lifestyle. DOABLE.
Best,
Jing
Jing Wang, PhD, MPH, RN
Associate Professor, Department of Nursing Systems, School of Nursing
Adjunct Associate Professor, School of Biomedical Informatics
Robert Wood Johnson Foundation Nurse Faculty Scholar (2013-2016)
University of Texas Health Science Center at Houston (UTHealth)
6901 Bertner Ave. | SON Room 580C| Houston, TX 77030
713 500 9022 tel | 713 500 2142 fax
Facilitator
Hi Jing,
Thanks for the comments and feedback – all great points. I agree that what’s outlined so far is too high-level and that we need to clearly define meaningful aims with demonstrable impacts; and present commercially viable methods and systems. I believe we already have the majority of what is required and will draft a more detailed solution architecture for discussion. Do you have time next Wednesday for a webex to review?
Ron