- In a path-breaking development that can change how radiology is delivered in Indian hospitals, qXR, an artificial intelligence (AI) tool to interpret chest X-rays, was developed by the Mumbai-based Qure.ai, a health-care technology company.
- Reading a chest X-ray is tough. The radiologists get it right only around 70-80% of the time.
- In a path-breaking development, qXR, an artificial intelligence (AI) tool to interpret chest X-rays, was developed by the Mumbai-based Qure.ai, a health-care technology company.
- The results of the trial were very promising in that the AI was calling the X-rays correctly around 90% of the time.
Overcoming the ‘black box’
- The most important feature of qXR was its ability to explain, in the way human radiologists do, why it interpreted the X-ray the way it did.
- Even the most sophisticated AIs frequently cannot do this — a problem known as AI’s black box.
- The black box is inherent in advanced AI techniques such as deep-learning.
- For example a neural network was trained to identify cancerous skin lesions as accurately as dermatologists could.
- Yet, when they asked the AI what part of a mole looked cancerous, it couldn’t answer.
- qXR is designed to avoid this problem and it is ground-breaking.
- It not only gives accurate results but can exactly tell how it arrived at the solution.
How does it work?
- To teach a computer to think like humans, researchers use a network of mathematical functions called as artificial neural network, which mimics the biological brain.
- Next, they input data into this network.
- In qXR’s case, these were chest X-rays and radiologist interpretations of them.
- When the network is exposed to millions of such X-rays and interpretations, it builds its own rules for translating the images into interpretations.
- The resulting AI can read new X-rays and spot abnormalities accurately.
- qXR was trained on over 1.5 million X-rays to detect 15 chest abnormalities, ranging from tuberculosis to potentially cancerous lung nodules.
- The AI’s findings were compared with the interpretations by three expert radiologists.
The Black box problem in AI
- The Black Box problem arises from the way we train our artificial intelligence systems.
- Most successful AI systems are trained using back-propagation.
- In this method the AI system learns by taking in inputs and corresponding outputs.
- It compares the actual output with the desired output and learns accordingly.
- This learning happens in a closed system called the internal state which is often ‘hidden’.
- However the problem is, it is not known how these systems learn since it is ‘hidden’. Therefore it is called a black box problem.
Impact on TB screening in India
- India has an estimated 2.7 million TB cases every year.
- One of the main problems is that many of these patients do not get a diagnosis early enough.
- TB screening is a bottleneck because they lack access to health care.
- Further if the doctors are not trained, they may fail to spot the disease in a chest X-ray.
- An AI that can distinguish likely TB cases from normal X-rays and send the likely patients for further testing can save radiologists a lot of time.
- As a result of accuracy of AI-based screening, only some 50 of every 1,000 healthy people screened are likely to need further testing.
- Therefore NITI Aayog has began talks with Qure.ai for a pilot tuberculosis (TB) screening project as part of its Aspirational Districts programme.
- However one of the main challenges in deploying qXR across rural India is lack of digitisation in hospitals.
- The problem is that none of the diagnostic centres digitise X-rays today, without which the AI can’t function.
Basics on Machine learning
- Machine learning is a branch of artificial intelligence that has been employed in a variety of applications to analyze complex data sets and find patterns and relationships among such data without being explicitly programmed.
- Machine learning algorithms analyze data features as inputs, and by the process of iterative improvement can produce linear and nonlinear predictive models.
- One of the types of machine learning algorithms is artificial neural networks (ANNs).
Applications of machine learning
- Machine learning has been used across many industries, including banking and finance, manufacturing, marketing, and telecommunications.
- Some more common every day examples include e-mail spam filters, face recognition, search engines, speech recognition, and language translation.