Predicting shortages
before they happen.

Enterprise pharmaceutical supply chain intelligence for India's 641 districts and 25,000 public health facilities.

Risk Intensity

LowHigh

Full-loop automation.

ICMR Translation Layer

Converts outbreaks to exact SKU volumes.

  • Ingests live Epidemic data streams
  • Maps standard ICMR treatment protocols
  • e.g., 1 Dengue case = specific IV fluid units

Adjacency Pre-Stocking

Proactively stocks before disease crosses borders.

  • Monitors neighbouring district outbreaks
  • Flags rising cross-border infection vectors
  • Auto-generates proactive Purchase Orders

Stock Rebalancing

Digitally redirects surplus inventory to high-risk zones.

  • Identifies overstocked health facilities
  • Re-routes surplus inventory dynamically
  • Eliminates waste and balances supply chains

Every signal. One dashboard.

Pharmasense ingests 6 real-time signals simultaneously, each powered by a distinct AI model — to detect outbreaks before health systems react.

ALERT
Epidemiological (IDSP)
↑ Dengue + Diarrhoeal Rising

80,370 cumulative dengue cases tracked across 641 districts. Diarrhoeal cases up 34% WoW.

Model: ST-GNN
HIGH RISK
Weather (IMD)
38°C · 82% humidity

Stagnant water in 3 districts. Peak mosquito breeding window activated.

Model: GraphWaveNet
ANOMALY
Search Trend Index
2.4× above baseline

'Dengue fever symptoms' spiked 2.4× above seasonal baseline in Delhi NCR.

Model: Prophet
MODERATE
Rainfall Index (IMD)
+40% of normal

Above-average monsoon onset. 3 districts with confirmed standing water.

Model: N-BEATS
ACTIVE
ML Ensemble Score
66% confidence

Ensemble predicts continued dengue + diarrhoeal co-rise over next 4 weeks.

Model: ST-GNN + TFT
WATCH
Inventory Stress
14 drugs <30d cover

14 disease-critical drugs have less than 30 days of supply cover remaining.

Model: LightGBM

7-model AI architecture.

Each model has a distinct role. No redundancy — the ensemble covers real-time, medium-term, and strategic horizons end-to-end.

ST-GNNSPATIAL
Spatio-Temporal Graph Neural Network

Real-time outbreak detection & spatial spread across districts

1–4 weeks
GraphWaveNetGRAPH
Adaptive Graph Convolution Network

Dynamic graph learning for weather + mobility signal correlation

1–3 weeks
TFTINTERPRETABLE
Temporal Fusion Transformer

Interpretable medium-term demand forecasting with 28 covariates

4–12 weeks
N-BEATSPURE DL
Neural Basis Expansion (NCD)

Non-communicable disease demand curves — pure DL, no covariates

2–8 weeks
Prophet
Meta Prophet — Strategic Planner
STRATEGIC

6–12 month demand forecasting with monsoon seasonality. Drives procurement budget planning.

3–12 months
LightGBM
Gradient Boosting Decision Trees
BOOSTING

Inventory stress scoring and drug criticality ranking using epidemiological + inventory features.

Continuous
Gemini 1.5
Google Gemini LLM
LLM

Translates model outputs into executive-readable briefs and purchase order justifications in natural language.

Real-time

Built for scale.

Pharmasense is designed to handle the scale of India's public health infrastructure. It continuously syncs with HMIS data to predict localized outbreaks and adjust stock requirements automatically.

  • Live Temporal Fusion Transformer (TFT) predictions
  • Automated inventory redistribution
  • Enterprise-grade Role Based Access

Real-time telemetry active

Built on trusted intelligence sources

HMIS DataIDSP Outbreak SignalsICMR Treatment GuidelinesIMD Weather TelemetryDVDMS InventoryNHM DirectivesHMIS DataIDSP Outbreak SignalsICMR Treatment GuidelinesIMD Weather TelemetryDVDMS InventoryNHM Directives

Ready to modernize?

Log in to the dashboard to view real-time risk assessments and pending automated purchase orders for your jurisdiction.