Segmenting and classifying tuberculosis, pneumonia and COVID-19 in chest X-rays
A deep learning pipeline that first isolates the lungs in a chest X-ray, then classifies the image as normal, tuberculosis, pneumonia or COVID-19.
Machine learning
Prediction, classification and forecasting models trained on your data and monitored in production.
Overview
Machine learning turns historical data into reliable predictions: who is at risk, what will sell, which claim looks unusual. We build models with sound validation, explain what drives each prediction, deploy them as APIs or dashboards, and monitor them so they keep performing after launch.
What we build
- Risk scoring
- Models that rank patients, borrowers, policies or assets by risk, with the factors behind each score.
- Forecasting
- Demand, yield, price and capacity forecasts that account for season, weather and local events.
- Anomaly and fraud detection
- Unsupervised and supervised methods that flag unusual transactions, claims or sensor readings for review.
- MLOps
- Versioned data and models, automated retraining, drift monitoring and alerting on your cloud.
Use cases by industry
Where machine learning makes the biggest difference for our clients.
Machine learning: case studies & projects
Filter all case studies
AI frameworks for cardiovascular and stroke risk in patients with comorbidities
A body of research on how AI can estimate cardiovascular and stroke risk in people with diabetes, Parkinson’s disease, rheumatoid arthritis, sleep apnoea and COVID-19.
Where AI saves money in healthcare: diagnosis versus treatment
An analysis of the economics of artificial intelligence across the healthcare pathway, recognised with MDPI Healthcare’s 2022 best paper award.
Machine learning and IoT for solar plant control and fault detection
A patented approach that combines IoT sensing with machine learning to control a solar plant, detect faults early and predict energy output.
Crop disease detection and advisory for a farmer producer organisation
Farmers photograph an affected plant; the app identifies the likely disease and gives treatment advice in Marathi, Hindi or English, while the FPO sees outbreaks on a map.
Farm digital twin for yield forecasting and irrigation planning
A field-level digital twin built from satellite imagery, weather and soil data that forecasts yield and recommends when and how much to irrigate.
Health insurance claims: document extraction and fraud signals
Automated reading of hospital bills, discharge summaries and prescriptions, with checks against policy terms and explained fraud signals for investigators.
Credit risk scoring and early-warning signals for an NBFC
Machine-learning credit scores with clear reason codes, plus an early-warning model that flags accounts likely to fall behind on payments.
Grievance classification, prioritisation and routing
Machine learning that reads incoming citizen grievances, assigns category and urgency, and routes each one to the right office, with dashboards for senior officers.
How we deliver
Discover
A short workshop to understand the problem, the data and what success looks like.
Prototype
A working proof of concept on a sample of your data, so you can judge results early.
Build
The full solution, tested against your data and integrated with your systems.
Deploy and support
Production deployment on your cloud, with monitoring and continuous improvement.
Questions
How much data do we need?
It depends on the problem. We start with a data review and tell you honestly whether your data can support a reliable model.
Can we see why the model made a prediction?
Yes. We use explainability methods such as SHAP so each prediction shows the factors that drove it.
Who runs the model after launch?
We can operate it for you or hand over to your team with documentation, monitoring dashboards and training.
More technologies
Explore machine learning for your organisation
Start with a free discovery call. We will tell you honestly what is possible with your data.