THE POTENTIAL OF AI IN SAVING LIVES:
Artificial Intelligence algorithms that are trained on medical records/data taken from actual patients from hospitals, actually have the potential to save lives by making a better diagnosis. But, in a lot of countries, it’s illegal to use data for such reasons. This limits the ability of AI systems to make predictions and diagnosis, as researchers only have small datasets to work with.
We know that AI and ML algorithms need vast amounts of data to be trained. But, we also know about the sensitivity of medical data to be leaked. That is the reason why, AI researchers are experimenting on ways that can help get the relevant medical data records to test their algorithms, trying really hard not to leak the data.
The Stanford Medical School in California is testing an AI system, with real patient records. Patients who willingly contribute their data to the system, help the AI diagnose eye disease, and refrain from getting into their personal details in the data.
Even though they have tried combining multiple technologies like Blockchain with AI to help the data stay immutable and secure, the data is still vulnerable to threats as the Blockchain applications are owned by other companies, and the threat is the vulnerability of the company, more than the application. If the company closes, for example, the data would be lost completely. So, the companies and researchers working together have to be vigilant all the time. But nevertheless, AI is still progressing every day and saving lives.
AI IN MENTAL HEALTH CARE:
Back in 2017, some researchers at Carnegie Mellon University and the University of Pittsburgh published a paper that discussed their recent AI experiment that could predict suicidal thoughts in individuals. They performed the experiment on 34 people, who were divided into groups. The team took scanned brain tests of all the individuals and the brain activity of suicidal individuals was very different from the others when words such as ‘death’ or ‘distressed’ or ‘troubled’ were presented to them. The ML algorithm was trained using these brain locations that showed changes, and the words presented. The results were phenomenal, 90% of individuals who had attempted suicide or had suicidal thoughts were tracked down. This research helped to get an idea of how a suicidal person thinks and acts accordingly. This early detection of suicidal thoughts gives the mental health workers or loved ones of the suicidal individuals to act before it’s too late.
An application named ‘The AI Buddy Project’ is one AI project that aims to help children whose parents are at war and the kids are struggling with their mental health. The technology provides a fictional companion or ‘avatar’ to the kids on their favourite digital platforms, which asks them questions, keeps a check on their behaviour, and forwards the progress reports to the guardians. The report is produced tracking the ‘words’ used in the conversations with the child. This helps to keep a close check on the mental well-being of the children.
AI IN ELECTRONIC HEALTH RECORDS OF PATIENTS:
Doctors might be able to save half of their workdays and give more time to the patients with the help of AI! Smart algorithms integrated with NLP, are being trained to identify the voices of patients with their questions and doctors with their answers, strategizing them all into electronic health records. This NLP- based system listen to the whole conversation between the doctor and patient, translates it, and strategically puts them into a medical report, after extracting the important information, phrases and keywords. That means, a doctor doesn’t have to put in extra efforts to record any sort of data of the patient, the smart algorithm handles all of it.
This same technology is all set to give a voice to the patients who don’t have one – like victims of traumatic brain injuries or strokes. But this technology will take almost a decade to come into the market. A lot of research and development is yet to be done in this arena.
AI IN RADIOLOGY AND DIAGNOSIS:
Lung cancer is deadly and is very hard to detect. Early detection of the tumour can help save millions of lives each year. AI is helping doctors all over the world to recognize primary nodules and classify them as malignant or benign. Not just this, but it is also helping in identifying and classifying liver lesions as benign or malignant.
AI can also help in detecting colonic polyps early on, that usually go undetected and can cause colorectal cancer with time. Screening a mammography report is hard to interpret, and often times identifying microcalcifications are extremely hard to interpret. AI can assist in their interpretation and characterization. AI is also being used to make diagnostic predictions of brain tumours.
Diagnosing lesions on the skin is a tough job to do. But Deep Learning algorithms inspect lesions and suspicious areas way better than humans, and more efficiently. Additionally, AI also accurately detects mitosis, do segmentation of histologic primitives, and classify tissues in pathology.
HOW DOES THE FUTURE LOOK LIKE?
AI algorithms scale with data. Which means, more data is generated each day, and with the continuous research, the predictions and performance of AI systems keep getting better. Medical imaging is an important pillar in medical treatments since the 1890s. Now, the advancements in imaging hardware and AI have enabled the discrimination in tissue’s densities. These differences are very hard to predict with a human eye. With time, the role of radiologists will broaden from a technical perspective too, and they will be expected to operate AI systems that will help them make a relevant diagnosis. As AI exceeds human intelligence and performance, and researches keep improving, we can expect it to augment the work medical personals do. So the human operators or medical personals will not only miss outcomes but will be able to interpret the reason behind any diagnosis.