How to Build a Remote Patient Monitoring Platform - IoT Devices, CMS Billing, Alert Engine, HIPAA Architecture 2026

Remote patient monitoring is the fastest-growing segment in digital health with a sustainable CMS reimbursement model that does not depend on venture capital. EngineerBabu built Somnoware, the remote monitoring platform for ResMed’s sleep therapy device network, subsequently acquired by ResMed. We understand production-grade medical IoT at scale.

2026 RPM billing codes:

Module 1 – FDA-Cleared Device Integration

RPM devices must be FDA-cleared. Consumer-grade fitness trackers do not qualify.

Device categories and integration methods:

The device abstraction layer:

A standardised internal interface normalises data from every device regardless of manufacturer format. Adding a new device requires implementing one new adapter not modifying the core platform.

Data model per reading:

The timestamp distinction matters for billing, CMS counts days where data was transmitted, not measured. Both timestamps must be recorded.

Module 2 – Real-Time Vital Signs Streaming and Alert Engine

The data pipeline:

Device → Bluetooth → patient smartphone → device manufacturer cloud → platform API → time-series database → alert evaluation

Time-series database choice:

Alert evaluation, every incoming reading is evaluated:

Alert fatigue prevention: four mechanisms:

1. Patient-specific thresholds: A systolic of 160 is a crisis for a 35-year-old, normal for an 82-year-old. Each patient gets provider-set thresholds.
2. Alert suppression windows: Expected post-meal glucose spikes do not generate alerts.
3. ML anomaly scoring: Deviation from the patient’s personal baseline, not population averages.
4. Alert deduplication: One alert for a sustained abnormal period, not one per reading.

Module 3 – CMS Billing Compliance Automation

The compliance tracking engine:

Mid-month compliance alerts (critical feature):

During the month not at month-end, the platform surfaces compliance gaps:

The billing automation workflow:

At month end, the platform automatically:

1. Calculates each patient’s transmission day count
2. Determines correct device supply code (99445 vs 99454)
3. Sums clinical management time per provider per patient
4. Determines correct clinical time code
5. Generates 837P claim files
6. Submits to clearinghouse
7. Tracks ERA responses and posts payments

Module 4 – Patient App and Engagement

The patient mobile app (iOS + Android, Flutter):

The engagement cadence:

Module 5 – Provider Dashboard and Care Coordinator Workflow

The provider dashboard:

The care coordinator workflow:

Cost to Build a Remote Patient Monitoring Platform

EngineerBabu built Somnoware, the RPM platform for ResMed’s sleep therapy device network, acquired by ResMed. Contact: mayank@engineerbabu.com

Frequently Asked Questions

• What is the 16-day rule in RPM billing and how does the platform track it?

CPT 99454 requires a minimum of 16 distinct calendar days of data transmission in a 30-day billing period. The 2026 update added CPT 99445 for 2 to 15 transmission days. The platform tracks transmission days automatically, counting any calendar day where at least one reading was received. At month-end, the platform selects the correct code based on the actual count. Mid-month alerts notify the care team when a patient is on track for 99445 instead of 99454, enabling proactive outreach to increase transmission frequency before the billing window closes.

• How did EngineerBabu’s Somnoware build inform this RPM architecture?

Somnoware was built for ResMed’s connected sleep therapy devices, CPAP and BiPAP machines that transmit usage data nightly. The core challenges were identical to a general RPM platform: device protocol normalisation across multiple device generations, alert engine design that surfaces clinically meaningful signals without alert fatigue, compliance tracking for billing purposes, and a patient app that drives engagement with patients who have chronic conditions. The acquisition by ResMed validated the platform’s clinical and technical quality. EngineerBabu applies these lessons, specifically the device abstraction layer architecture, the alert fatigue prevention model, and the compliance billing engine to new RPM builds.