Is that cough serious?

Helfie’s Cough AI uses clinically studied cough-analysis technology to screen for potential signs of respiratory conditions, including tuberculosis, COVID-19, COPD, and asthma.

• Accuracy
  • Tuberculosis
  • 86.82% accuracy
  • 90.36% sensitivity
  • COVID-19
  • 75.54% accuracy
  • 95.45% sensitivity
  • COPD
  • 86.3% accuracy
  • 61.1% sensitivity
  • Asthma
  • 83.0% accuracy
  • 68.2% sensitivity
    Accuracy measures overall correct classifications.
    Sensitivity measures the proportion of people with the condition correctly identified.
• Algorithm
  • Mel-frequency cepstral coefficients (MFCC) for acoustic feature extraction.
  • Convolutional neural networks (CNNs) and feedforward neural networks for tuberculosis and COVID-19.
  • Audio Spectrogram Transformer (AST) for COPD and asthma.
• Capture mechanism
  • Non-invasive cough recording via smartphone microphone.
• Training data
  • Thousands of cough recordings collected and analysed across clinical research studies.
• Research and clinical validation
  • Powered by Swaasa (Salcit Technologies), Google Health Acoustic Representations (HeAR), EPFL Switzerland, and Helfie R&D Labs.
  • Supported by peer-reviewed research and additional clinical studies involving thousands of participants.
  • Clinical research institutions include Apollo Hospitals, Government TB and Chest Hospital, and Christian Medical College (CMC) Vellore.

Your Health Intelligence (HI) connects information from your cough checks, conversations, and uploads to build a clearer picture of your respiratory health over time.

By tracking results and identifying patterns, it can help you notice potential changes in your respiratory health and understand how factors such as lifestyle, environment, and existing conditions may be affecting it.

Helfie’s Health Intelligence (HI) brings together information from your health checks, conversations, uploads, and history, then connects it with a vast body of medical knowledge to build an increasingly informed picture of your health.

Rather than relying on a single AI model or looking at each result in isolation, HI uses multiple layers of intelligence and machine learning to interpret health signals, connect related information across different areas of your health, and understand those results in the context of you.

It also maintains an active memory of your health journey, so previous results, patterns, and interactions can inform what comes next.

The result is intelligence that is personal, contextual, and designed to help you better understand and manage your health over time.

A health check gives your Health Intelligence new information about you. You can then talk to your HI about those results, ask what they mean, explore possible reasons for changes, and understand them in the context of previous checks and health history.

The conversation also becomes part of that context. So when you check again later, your HI can consider both your new results and what it’s learned from earlier checks and conversations.

This is how it builds a more detailed, connected picture of your health over time.

Helfie makes it easy to share your Cough AI results and respiratory health trends with your doctor or healthcare provider, giving them additional information to support their assessment.

Sharing results from multiple checks, alongside your symptoms and health history, can help your healthcare provider understand changes over time and make better-informed decisions.

To improve accuracy, Helfie’s Cough AI needs a few key details to refine your assessment.

Currently, Helfie uses the following inputs for this check: Age, Location and Symptoms.