Hello, I will be speaking for AI in Healthcare only.

The major adaptation areas in health care are Diagnostics, Research, and to some extent pharmaceutical industry for pharmacoKinetics.

Diagnostic is the biggest area currently undertaken using computer vision to interpret MRI, CT-Scan, X-Rays, Ultrasonic evaluations, Eye inspection, Skin inspection, and hematological analysis, to name a few.

I have worked for 2 of the use cases, cervical cytology detection with pap slides, and liver cancer detection with hepatocellular biopsy slides.

The major challenge faced was acquiring annotated/segmented image data. While pap slides were easy to acquire from some open data sources, the hepatocellular were hard to obtain, and a pathologist was contracted for the same, while ideally, a group of pathologists should have annotated the slides to remove any sort of bias.

Also, the models are mainly assisting the pathologists as they may not be used standalone due absence of any regulatory framework per my knowledge.

There was an article published on AI use with dermatologists, where the AI was trained on different moles to detect a variety of cancers and other diseases. The study had doctors work with AI programs to identify diseases, and then checked the comparative success rate of each.

Interestingly, the AI excelled at identifying rare diseases, while the doctor was far more successful at identifying common diseases and cancers.

The article suggests that instead of replacing humans with AI, coordinating the two, with AI focusing on rare diseases that most doctors do not see often in practice while the Doctors focus on more common diseases may prove effective.

This provides further support to the broader belief that AI and humans working together is more effective than either alone.

The greatest challenge in AI in general is the lack of abundant quality data.

Diagnosis and Treatment Advances: AI algorithms can analyze medical data, including imaging scans and patient records, to help with disease diagnosis and create individualized treatment programs.

Predictive analytics: AI makes it possible to forecast patient outcomes and identify high-risk patients, enabling healthcare professionals to take early action and stop negative outcomes.

medication Discovery: AI algorithms are being used to speed up the process of finding new drugs by analyzing enormous volumes of data and identifying possible medication candidates.

Challenges:

Data Security and Privacy: Protecting patient privacy and data security is essential when adopting AI systems because healthcare data is extremely sensitive.

Ethics: There are questions of transparency, accountability, and AI decision-making that need to be answered. Responsible AI procedures and rules are required.

Another use case is gnomic mapping which has been underway in Britain, paving the way for stratified health care based on genome mapping and health records of individuals and prescribing the custom medicinal course and dosage to reduce side effects caused by general medicinal administration.

I want to comment on the challenge perspective. Accuracy and precision which will be based on the data quality used for training purposes. Standards and methods to ensure this critical point is critical.

The integration of AI, blockchain, and the Internet of Things (IoT) in the healthcare sector has the potential to revolutionize healthcare delivery, improve patient outcomes, and enhance operational efficiency. Here are some recent advances and challenges in implementing these technologies:

Advances:

Challenges:

AI Assessment Tools: AI assessment tools aim to evaluate the performance, fairness, and safety of AI systems. They help assess biases, model accuracy, robustness, and compliance with ethical guidelines. Some popular AI assessment tools include IBM AI Fairness 360, Google's What-If Tool, and Microsoft's InterpretML. These tools assist in evaluating AI models, identifying potential issues, and facilitating the development of responsible AI applications.