A quantitative model to ensure capacity sufficient for timely access to care in a remote patient monitoring program

Abstract Introduction Algorithm‐enabled remote patient monitoring (RPM) programs pose novel operational challenges. For clinics developing and deploying such programs, no standardized model is available to ensure capacity sufficient for timely access to care. We developed a flexible model and interactive dashboard of capacity planning for whole‐population RPM‐based care for T1D. Methods Data were gathered from a weekly RPM program for 277 paediatric patients with T1D at a paediatric academic medical centre. Through the analysis of 2 years of observational operational data and iterative interviews with the care team, we identified the primary operational, population, and workforce metrics that drive demand for care providers. Based on these metrics, an interactive model was designed to facilitate capacity planning and deployed as a dashboard. Results The primary population‐level drivers of demand are the number of patients in the program, the rate at which patients enrol and graduate from the program, and the average frequency at which patients require a review of their data. The primary modifiable clinic‐level drivers of capacity are the number of care providers, the time required to review patient data and contact a patient, and the number of hours each provider allocates to the program each week. At the institution studied, the model identified a variety of practical operational approaches to better match the demand for patient care. Conclusion We designed a generalizable, systematic model for capacity planning for a paediatric endocrinology clinic providing RPM for T1D. We deployed this model as an interactive dashboard and used it to facilitate expansion of a novel care program (4 T Study) for newly diagnosed patients with T1D. This model may facilitate the systematic design of RPM‐based care programs.

companion tool to the financial model previously developed by our team to aid in the management of telemedicine-based diabetes care clinics. 14 Here, we describe the creation of a model to aid in capacity planning and an interactive dashboard as a specific instance of this model. We sought to design and evaluate a formal capacity planning model in order to reduce barriers to implementing novel interventions such as RPM.

| ME THODS
Setting: This study took place at a paediatric T1D clinic at an academic medical centre as part of the Teamwork, Targets, Technology and Tight Control (4 T) Study and the CGM Time in Range Program at Stanford (CGM TIPS), programs that provide RPM based on data from CGM. Both programs were approved by The Stanford University Institutional Review Board and participants or their legal guardians gave informed consent. The details of the study protocols, the patient populations and the development of the technology used have been reported in detail. 15,16 All participants in the 4 T Study and its predecessor, the 4 T Pilot Study, were started on a CGM system (Dexcom G6, Dexcom Inc) within the first month of diabetes diagnosis. 16,17 Participants in the 4 T Pilot and the 4 T Study 1 were between the ages of 1 and 21, were newly diagnosed with T1D over the past 30 days and were willing to operate an Apple device whose data was shared with the T1D clinic. Weekly remote monitoring of CGM data was provided for newly diagnosed T1D patients beginning in March 2019. Participants in the CGM TIPS study were enrolled between July 2020 and April 2022, and the study is currently capacity planning, diabetes technology, paediatric endocrinology, remote patient monitoring, type 1 diabetes ongoing. 18,19 The number of participants continues to grow with enrolment throughout the study. These participants were covered by public insurance, diagnosed with T1D of any duration and were willing to operate an Apple device whose data was shared with the T1D clinic.
The RPM program has four primary components: (1) each patient wears a CGM from which data are uploaded to the manufacturer's server via the patient's mobile device from which they are available to be downloaded by the clinic, (2) CDCESs review CGM data weekly for the 4 T population and monthly for the TIPS population, (3) a platform automatically downloads the data and flags patients for review based on a combination of American Diabetes Association consensus metrics and algorithms, 7,9,15 and (4) CDCESs review flagged patient data, send patients secure messages through the electronic medical record (EMR), and update the patient chart in the EMR.
General setting: The capacity planning model avoids assumptions specific to the use of continuous glucose monitors or the workflows or population of the study institution. It is designed for any RPM program that has the same four primary components: (1) a population of patients about whom data are available remotely to care providers, (2) a fixed cadence at which care providers dedicate time to RPM-based patient care, for example 1 h every Friday, (3) a tool that analyses patient data to flag some of the patients for review by care providers and (4) and care providers (e.g. a CDCES or physician) that review the flagged patient data and based on the data select patients to contact to provide guidance.
In the TIDE program, flagged patients are those identified by the algorithm as potentially needing attention due to certain criteria being met, such as spending more than 5% of the time with glucose levels below 70 mg/dL. The criteria for review and outreach may vary depending on the study or clinic's specific guidelines. In our experience, a small subset of flagged patients actually end up not requiring contact because, upon review, the provider may determine that no immediate action is needed or that the patient is already on an appropriate course of action. This is an important aspect for clinics to consider when estimating the percentage of patients requiring contact, as it may significantly affect the model's estimates.
To limit the number of contacts that patients receive and the amount of provider time required, the thresholds for patient contact can be set based on historical patient data in order to limit the percentage of patients flagged to an appropriate level. 10 For clinics with significantly higher capacity or populations with different needs or preferences, a higher percentage of patients may be contacted each review period, and this model allows the clinic to plan accordingly. in these three studies, whose demographics are depicted in Table 1.  The number of patients identified for contact was defined as the percentage of patients who need contact multiplied by the number of patients in the program. The CDCES capacity was defined as the number of minutes per review period that each CDCES had available to review patient data and send messages to patients.

Remote monitoring was conducted on a weekly basis for 4 T and
Flagged patients are those identified by the algorithm as potentially needing attention due to certain criteria being met (e.g., glucose levels outside a specified range). However, not all flagged patients will ultimately require contact, as a provider may review the data and determine that no immediate action is needed. In our experience, the distinction between flagged patients and those requiring contact is minimal, as our model is designed to accurately identify patients in need of intervention. The demand was measured as minutes required for CDCESs to review patient data and send messages to those patients who meet criteria for contact.
The primary output was the calculated capacity of the care team as a percentage of the time required to review and contact all of the patients identified as needing review and contact. A secondary output was a table of potential modifications, to care provider capacity or the patient population, sufficient to ensure that the calculated capacity meets or exceeds patient demand.

| RE SULTS
The dashboard provides an interface for users to compare their clinic capacity with patient need by varying input parameters and seeing model outputs update dynamically ( Figure 3).

| DISCUSS ION
We designed a generalizable, interactive model for capacity planning that facilitated improved resource allocation to maximize the efficiency of a paediatric endocrinology clinic providing RPM for T1D. The model facilitated the quantitative design of the expansion of RPM to additional patients by identifying the necessary resources.
This model has the potential to improve demand-capacity matching and facilitate efficient care delivery for clinics using or designing RPM programs for diabetes care. One specific application will be in the dissemination of the 4 T Study to other diabetes centres.
We developed a proof-of-concept interactive dashboard for capacity planning, which provides a platform to help stakeholders make better operational decisions for population-level diabetes care. The deployment of the model as an interactive dashboard and its use by clinical leaders revealed several insights. We found that reductions to average patient review time, increases in the number of

CDCESs and increases in a CDCES's available hours per time period
through the standardization of workflows has the potential to produce substantial improvements to clinic capacity. Hiring additional  It is important to note that the entire concept is based on timeeffective asynchronous consultation, where a physician or diabetes expert evaluates marked reports and changes therapy, putting the information into an electronic patient record. No more timeconsuming phone call or video call is covered here. For true staff scheduling, it would be necessary to calculate those times as well.
We will consider adding this aspect to the discussion. Finally, our model is limited because it was calibrated using 1 month of data and tracks the specific characteristics of the patient population in our clinic. However, this also highlights an opportunity in providing a template of building a dashboard based on past data at a single institution that would become more robust with data from more clinics. The scenario analysis from the interactive dashboard is also expected to become stronger with data over longer periods of time.
We acknowledge that advisory software, which even makes suggestions for therapy (pump/MDI) settings, could be the next step in the evolution of our tool. Although such software is not yet available in most countries, the workload might remain the same. Children will use more and more AID pumps in the future, and the data evaluation is much more complex (depending on the software). The more complex the therapy, the more synchronous and thus more timeconsuming, face-to-face counselling has to be done. Thus, our tool F I G U R E 4 Number of patients shown (green), flagged for review (red), and contacted (blue) from 2020 to 2022 in the TIDE Program. These historical data metrics were compared with dashboard projections to assess the capacity planning dashboard and identify areas for operational improvements.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the senior author, DS. The data are not publicly available due to restrictions that could compromise the privacy of research participants in the remote monitoring program. The model is publicly online: https://surftide.shinyapps.io/capacitydashboard/.

CO N S E NT
The Stanford Institutional Review Board approved this protocol and consent (and assent for participants aged 7-18 years old) was obtained for review of all participants.