Statement of Work.pdf

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG 1

Statement of Work (SOW)

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG: Automated Signal Analysis, and Clinical Correlation

1. Background and Purpose

• Wearables enable continuous, non-invasive monitoring. PPG offers insights into arterial compliance, vascular age, arrhythmia detection, endothelial function, and HRV, but real-world fidelity/robustness limits adoption.

• Speckle-plethysmography (SPG) measures red blood cell movement as a blood flow index, producing a waveform similar to but distinct from PPG. Combining SPG with PPG may improve non-invasive blood pressure and other cardiovascular assessments.

• There is a need for automated SPG waveform analysis, and signal processing expertise to translate SPG/PPG into clinically useful outputs.

2. Project Objectives The Contractor shall deliver analytical outcomes that:

• Demonstrate that SPG waveform features correlate with established markers of cardiovascular function and resemble finger arterial pressure morphology.

• Show improved performance of SPG+PPG models versus PPG-only for noninvasive blood pressure and related endpoints.

• Provide a mechanistic interpretation of SPG signal origins with respect to cardiovascular physiology.

This effort is project-based. The Government is procuring defined outcomes and deliverables, not individual labor hours, staffing models, or specific personnel.

3. Scope of Work

• Develop and validate automated signal processing for SPG and combined SPG+PPG signals.

• Quantify relationships between SPG features and known cardiovascular markers;

evaluate improvements over PPG-only approaches.

• Deliver software, documentation, and reports suitable for internal NIH research use and potential real-time deployment.

4. Tasks and Deliverables

Task 1: Project Plan and Data Strategy o Define data sources, preprocessing, labeling, storage, and protection; outline statistical analysis plans.

o Identify risks (e.g., motion artifacts) and mitigation.

Task 2: Signal Processing Pipeline o Build automated, modular pipelines for SPG and PPG preprocessing, artifact rejection, feature extraction, and quality indexing.

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG 2 o Provide configurable real-time and offline modes.

Task 3: Model Development and Validation o Train and evaluate regression/classification models for BP and other targets using SPG, PPG, and combined inputs.

o Perform cross-validation and holdout testing; compare predictive value of SPG to

PPG.

Task 4: Clinical Correlation Analyses o Relate SPG parameters to known cardiovascular markers (e.g., pulse wave morphology indices, HRV metrics, endothelial function proxies).

o Provide mechanistic interpretation supported by literature and empirical results.

Task 5: Real-Time Demonstration and Tech Transfer o Deliver a real-time demonstration package and complete technology transfer

(code, documentation, test datasets, and user guides).

o

5. Deliverables, Due Dates, and Acceptance Criteria D1. Project Plan (30 days ARO) o Contents: work breakdown, schedule, data management plan, analysis plan, risk register.

o Acceptance: Government review/approval within 10 business days; revisions as needed.

D2. Signal Processing Pipeline v1 (by end of Month 3) o Contents: source code (e.g., Python/MATLAB), unit tests, API docs, and a test report on Government-provided datasets.

o Acceptance: Reproducible on reference environment; achieves predefined artifact suppression and SNR improvement on test set; passes provided unit tests; code quality and documentation meet agency standards.

D3. Feature Set and Clinical Correlation Report (by end of Month 6) o Contents: feature catalog; methods; stats; correlations with confidence intervals;

dataset description; limitations.

o Acceptance: Methods/statistics reproducible; for at least three predefined markers, observe meaningful associations (e.g., r ≥ 0.6 or equivalent effect sizes on agreed datasets) or provide analysis explaining divergence.

D4.SPG-only, PPG-only, and SPG+PPG estimation of BP (by end of Month 9) o Contents: training scripts; evaluation on holdout set; Bland-Altman and

MAE/RMSE metrics.

o Acceptance: Quantification of the performance of SPG+PPG vs PPG only on agreed metrics and dataset; example thresholds for research use: MAE at or below 7 mmHg systolic and 5 mmHg diastolic on the holdout set, with limits of agreement within ±15 mmHg; results reproducible by Government.

D5. Real-Time Demonstration Package (by end of Month 11) o Contents: executable or container; sample data stream; performance logs; user guide; demonstration video or live demo.

o Acceptance: Generation of technical system performance report evaluated by the agency; example goals: End-to-end latency <100 ms at ≥ 60 Hz sampling on

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG 3 reference hardware; memory/CPU within agreed constraints; passes scenario tests with simulated motion artifacts.

D6. Final Report and Tech Transfer (by end of Month 12) o Contents: comprehensive report; code repository handoff; model cards;

configuration files; de-identified test datasets or data generation scripts; future work and limitations.

o Acceptance: All materials complete and sufficient for independent replication by Government; issues log resolved or dispositioned.

o

6. Period of Performance 12 months from award.

7. Place of Performance Contractor facilities, remote performance permitted.

8. Government-Furnished Information/Property De-identified datasets (if available), data dictionaries, any applicable internal test protocols, and a reference computing environment specification.

9. Contractor-Furnished Items All personnel, software tools, computing environments, cloud resources, and licenses necessary to perform.

10. Security, Privacy, and Compliance

• Work will use only de-identified data; no PHI. De-identification will be performed by a designated individual within the government. This individual will be solely responsible for generating and retaining the re-identification key, which will be stored securely and not shared. All data provided to the contractor will contain pseudonymous identifiers, no personal identifiers, and only the minimum necessary demographic variables required for the analysis(age, sex, cohort etc). A second designated individual within the organization will ensure correct de-identification before release of data to the contractor. All study references will be removed. If it is later determined that PII/PHI is necessary for the performance of this work, Contractor must obtain prior written approval and comply with HIPAA and HHS regulations.

• In order to access NIH cloud-based systems storing de-identified data, contractor is required to complete the SOW, contractor must meet HHS/NIH security and training requirements; any cloud use must be FedRAMP Moderate (or higher) for CUI.

• Secure Coding and software development will adhere to contractual requirements.

• Comply with applicable export control laws; no ITAR-controlled work.

11. Intellectual Property and Data Rights

• Government obtains Government Purpose Rights to all deliverables, including source code, models, and documentation, unlimited rights in reports. Background IP remains

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG 4 with the Contractor; any third-party code must be disclosed with license terms compatible with Government use.

• If open sourcing is contemplated, the Government will approve license terms in advance.

Advancing Robust Optical Cardiovascular Monitoring via SPG and PPG 5

12. Reporting and Meetings

• Biweekly progress updates by email; monthly progress call; quarterly written status reports covering accomplishments, metrics, risks, and plans.

• Kickoff within 10 business days of award; final closeout meeting within 10 business days of final delivery.

13. Standards and References

• Follow good software engineering practices (version control, CI, unit tests, code review).

• Use appropriate evaluation standards/guidelines for cuffless BP research (e.g., ISO

81060-2 for reference device context; transparent model reporting).

14. Travel Not anticipated; any travel must be pre-approved and will be in accordance with the contract.

15. Quality Assurance Government will apply a Quality Assurance Surveillance Plan aligned to the acceptance criteria above (schedule adherence, reproducibility, accuracy metrics, documentation quality, responsiveness).