What’s new in Medicine?
You’re missing out on new developments in Medicine.
AI for Medicine specialisation offered by deeplearning.ai is, in my opinion, designed for software engineers, machine learning practitioners, and data scientists. The prerequisites for this specialisation are Statistics and Computer Science, particularly programming with Python.
The specialisation is structured into three courses:
- AI for Medical Diagnosis
- AI for Medical Prognosis
- AI for Medical Treatment
Here, I will not discuss in detail the content of the three courses. But I will highlight two topics I find insightful.
Topic 1 of 2: Medical Image Diagnosis with Computer Vision
AI is good (comparable to experts’ diagnosis) for medical image diagnosis, particularly for the below examples.
For medical image diagnosis to work well, we need to train computers to identify medical images. For example, pneumonia or no pneumonia. To do so, we need to label existing medical images with words such as Image_ID_1 = Pneumonia, Image_ID_2 = No_Pneumonia, Image_ID_N so on… After which, for computers to perform well, we need to train the computer with loads of medical images and labels. So with loads of images, it will be too laborious for humans to label each image using radiology reports. Therefore, we need to automatically label each image.
Topic 2 of 2: Medical Image Labelling with Natural Language Processing
Automated medical image diagnosis and labelling are simply predictions. Thus, we need to recognise that there could be false positives predictions. Therefore, we need to evaluate our prediction models carefully. By evaluation, for example, we need to decide if we want to accept a 10% error rate in prediction, and this is another topic for discussion. The AI for Medicine specialisation covers topics on how to evaluate a model and the key metrics to look out for.
All in all, this is an insightful specialisation with plenty of meat packed in it. I highly recommend AI for Medicine specialisation to my readers. Notwithstanding, this specialisation showed me the enormous opportunities that AI can bring to healthcare, such as cheaper and faster melanoma detection; AI can relieve radiologists from writing radiology reports while allowing radiologists to focus on checking or correcting the computer generated reports. I am looking forward to the changes AI will bring to many other industries.
