ABDM-Aligned Deterministic Metabolic CDSS at a digital metabolic clinic, Durg

Running a busy metabolic OPD in Durg—with 40 to 60 patients daily—presented several recurring operational and clinical challenges.

Calculating basal metabolic rates and renal dose adjustments manually during time-constrained consultations was inherently error-prone, with approximately 10% of calculations resulting in arithmetic errors.

Keeping up with the rapidly evolving diabetes literature also consumed approximately 10 hours per week.

Furthermore, the predominantly Hindi-speaking patient population was often provided with complex English diet sheets, resulting in poor understanding and adherence.

A digital solution was therefore essential. While standard EMRs supported basic clinical documentation, commercially available AI calculators produced variable and inconsistent outputs and lacked the deterministic mathematical safety required for precise metabolic dosing.

The stakeholders most affected were the clinician, who experienced significant cognitive and administrative burden, and patients, who experienced delayed retinopathy screening and low adherence to therapeutic recommendations.

The challenge directly impacted key NABH quality aspects, including:

• Clinical Care Quality: Standardization of clinical protocols and calculations.
• Patient Safety: Accurate risk prediction and reliable dosing.
• Patient Education: Localized, comprehensible, and patient-centred therapeutic interventions.

The Metabolic Command Center operates as a dual-layer hybrid digital ecosystem designed to combine secure patient connectivity with deterministic clinical computation.

Layer 1: Patient Communication Pipeline

The patient communication layer utilizes Eka Care, a secure, ABDM-compliant commercial EMR, to support:

• Seamless Ayushman Bharat Health Account (ABHA) ID generation.
• Pre-consultation health assessments.
• Secure WhatsApp-based Remote Patient Monitoring (RPM).
• Glucometer data uploads and remote monitoring.

This layer enables efficient communication and digital engagement with patients throughout the care journey.

Layer 2: Clinical Computation Engine

For complex metabolic calculations where commercial AI tools may not provide deterministic outputs, a custom-built and locally hosted Python/Streamlit application was developed.

The in-house clinical computation engine functions as a deterministic physiological calculator, ensuring that:

• Basal metabolic calculations are standardized.
• Renal dose adjustments follow defined computational logic.
• Calculations are logged and reproducible.
• Clinical computation remains consistent and auditable.
• Patient health information (PHI) is not exposed outside the controlled environment.

Hybrid Digital Ecosystem

This hybrid approach combines enterprise-grade patient connectivity with highly controlled, proprietary clinical decision-support capabilities.

The patient communication infrastructure supports the digital delivery of more than 5,000 prescriptions along with localized Hindi health and dietary infographics.

By integrating secure patient engagement, remote monitoring, deterministic clinical calculations, and localized communication, the Metabolic Command Center improves point-of-care efficiency, standardizes clinical computations, and supports better patient understanding and adherence.

Implementation Approach

The ecosystem was not introduced as a planned corporate rollout; rather, it evolved organically over a 12-month period through iterative development alongside a full-time clinical practice. The solution was developed to address immediate point-of-care challenges and was continuously refined based on real-world clinical requirements.

As the sole internal champion and clinical developer, I managed the end-to-end design and development of the clinical computation layer, including architectural design, Python coding, and clinical validation.

Because the complex informatics processes operate in the background of the consultation workflow, staff training requirements were targeted and minimally disruptive.

Reception and administrative staff were trained specifically on the use of the Eka Care platform, with emphasis on seamless ABHA ID enrollment and related patient-intake protocols.

Replicability

The model demonstrates that a hybrid digital ecosystem—combining a commercial EMR with a custom, locally hosted clinical computation engine—can be implemented within a single-physician practice in Tier-2 India without dedicated institutional IT support.

The approach provides a potentially replicable framework for clinicians and smaller healthcare practices seeking to introduce focused digital innovation using commercially available infrastructure combined with locally developed, clinically validated tools.

Operational Improvements

The implementation significantly transformed the clinical workflow. Weekly literature review time was reduced from approximately 10 hours to under 30 minutes, enabling more efficient access to relevant clinical information.

Diet chart generation time decreased from 8–10 minutes per patient to less than 2 minutes, recovering several hours of clinical capacity each week for direct, face-to-face patient examination and care.

Quality and Patient Safety Improvements

Manual calculation errors in metabolic prescriptions were reduced to zero within the implemented workflow.

Every computation is now digitally logged, reproducible, and audit-ready, strengthening traceability and supporting standardized clinical decision-making.

The ecosystem also enabled more than 400 point-of-care retinopathy screenings to be performed without waiting or pharmacological dilation, supporting earlier identification of retinal findings.

Patient Engagement and Education

Localized Hindi infographics and prescription-related guidance were delivered to patients through WhatsApp via the EMR platform. This converted complex therapeutic recommendations into more accessible, actionable routines for patients.

The digital communication ecosystem engaged more than 5,000 active patients, supporting ongoing patient education, communication, and adherence.

NABH Digital Health Alignment

The ecosystem directly supports the organization’s digital health objectives by establishing documented and metric-driven processes across:

• Health information management
• Medication safety
• Patient education
• Clinical decision support
• Continuous care tracking
• Digital documentation and auditability

Overall, the initiative demonstrates how a clinician-led digital ecosystem can improve operational efficiency, strengthen traceability and patient safety, and support continuous, data-driven healthcare delivery.

Key Enabler

The primary enabler was the strategic adoption of a hybrid digital architecture.

Rather than developing a patient communication platform from scratch, the initiative leveraged Eka Care EMR for secure, ABDM-compliant WhatsApp-based patient engagement and communication.

For clinical decision support, however, commercially available AI tools produced variable and inconsistent outputs. To address this limitation, a proprietary clinical computation engine was developed and hosted locally using Python and Streamlit.

Core Innovation

The core innovation lies in establishing a clear architectural boundary between patient communication and critical clinical computation:

• Enterprise EMR infrastructure is used for secure communication, patient engagement, and information delivery.
• Critical clinical calculations are performed through locally hosted, deterministic algorithms with standardized computational logic.

This approach was adopted as a patient-safety measure. For clinical applications such as metabolic dose calculation, reproducibility and consistency of computational outputs are essential; variable responses to identical inputs would be inappropriate for such high-stakes calculations.

Challenges and How They Were Addressed

The most significant challenge was time and development capacity. Building and validating the Python-based computation ecosystem alongside a full-time clinical practice required substantial effort, much of which was undertaken outside regular clinic hours.

The integration of the locally developed computation engine with the EMR’s established WhatsApp delivery pipeline created a practical digital workflow that combined controlled clinical computation with efficient patient communication.

This hybrid architecture ultimately functioned as a force multiplier, allowing clinical decision-support capabilities and patient engagement infrastructure to operate together while maintaining a clear distinction between communication technology and safety-critical clinical computation.

Top Learning

The key learning from this initiative is that the primary barrier to digital health innovation in Tier-2 India is often motivation rather than technical capability. Accessible technologies such as Python, Streamlit, and PubMed APIs can enable meaningful digital solutions without requiring extensive technology infrastructure. The critical requirement is a clinician willing to identify a practical problem and work iteratively toward a solution.

Recommendations for Other Hospitals

Hospitals seeking to implement similar digital innovations should:

• Begin with the single most painful or time-consuming part of the clinical workflow.
• Automate that specific process before expanding the scope.
• Avoid attempting to build a complex, multi-layer digital ecosystem from the outset.
• Develop the solution incrementally and layer additional capabilities sequentially based on demonstrated clinical need and user experience.

Sustainability

Sustainability is incorporated into the architecture of the solution.

The clinical computation engine operates on local hardware, while a stable commercial EMR is used for front-end patient communication. This architecture minimizes infrastructure requirements and avoids the need for dedicated IT personnel for routine maintenance.

The ongoing costs are therefore primarily limited to standard EMR and API usage, with no separate recurring infrastructure requirement for the locally hosted computation layer.

Supporting Visuals

Supporting visuals uploaded for the initiative include:

1. Eka Care WhatsApp Patient Engagement Flow
2. Local Python Clinical Computation Architecture Schematic