By Francis Han, Senior Director - Oracle Solution Center
Artificial Intelligence (AI) is increasingly making a huge impact on our lives and the current pandemic is accelerating the pace. This has been made possible with various contributing factors like the advancement of semiconductor technologies resulting in more powerful CPUs and GPUs, internet speed, rate of adoption of digital technologies, and improvements in pattern recognition methods. Pattern recognition (and the algorithms that support it) is not a recent phenomenon. In fact, Linear Regression (one of the most familiar algorithms) dates back to 1805. Through the many decades, and in between several dark winters, the evolution of more advanced machine learning and deep learning algorithms have brought us to a better place today.
Chatbots have been deployed in various healthcare solutions wherever there is a need to bridge the language gap between humans and machines. One way to make machines understand our human language is through Natural Language Understanding (NLU). Some of the key elements of NLU are the Intents and Entities. Let’s look at an example question (utterance) that a machine receives from a chatbot user: “What is the weather in London now?”. Here, the Intent is “Weather”, and there are two entities: “Place” and “Time”. If any required entity is missing, the chatbot is typically programmed to ask for it. That gives you an idea. Besides NLU, chatbots are equipped with NLP (Natural Language Processing). NLP helps the machine to ascertain what was said and NLU helps the machine to look at what was meant. Together with other components, the NLP and NLU help the chatbot to communicate with users using the human natural language. This enable chatbots to talk to humans, understand their questions, ask them questions and make inferences about their conditions. Such chatbots could also pass on relevant information to healthcare workers who would then take appropriate actions. Chatbots can improve the mood of humans, check on their conditions, follow up on their medications, etc. Further, incorporating the chatbots in Telemedicine and Assisted Living use cases can create immense benefits.
Digital Twins deployed in modeling humans could help speed up effective treatment for the individual. Fifty years ago, a digital twin environment was used to help to save the lives of astronauts on Apollo 13. In roughly the same way, a virtual model of the human body complete with clinical data, genomic information and AI algorithms can be leveraged to personalize medication for humans, to predict the outcome of specific procedures as well as to manage chronic diseases. The Digital Twin can also be used to model the hospital, together with its infrastructure, human resources and clinical flows. The objective would be to improve patient experience, clinical outcomes, caregiver experience and to lower costs by optimizing the use of resources.
Telemedicine as its roots in the early years of space program. Prior to Yuri Gagarin’s first flight into space, there were questions about whether the absence of gravity could have any adverse effect on his circulatory and respiratory functions. Hence the start of monitoring of animals in space with the biometric data sent to scientists on earth via telemetric link. This was followed by tests on other physiological or psychological effects on the human body. Today telemedicine is deployed to provide better diagnosis, recommend treatments, and giving round-the-clock assistance to the elderly. With telemedicine, patient visits can be reduced hence lightening doctors’ workloads. Doctors can monitor, diagnose, and treat diseases remotely thanks to telemedicine. AI technologies that are helpful for telemedicine include areas like NLP/NLU, Pattern Recognition leveraging the respective algorithms, Computer Vision using Deep Learning to decipher complex but identifiable patterns.
Disease Prediction is one area that can create profound implications in the field of medicine, for example in Telemedicine. The key here is early detection so that suspect cases could be referred to specialists for further review, enabling the patient to seek medical help early before disaster strikes. This applies to many forms of diseases, and the one thing that makes it possible is data. Data coming from the respective scanners and biometric devices provide clues about underlying problems. Various Unsupervised Learning as well as Deep Learning algorithms have been deployed depending on actual requirements. Models are trained to recognize anomalies and help us assess if the patterns point to possibilities of certain diseases. This means that the limited pool of medical specialists and related resources could focus more on treatment and less on screening.
Artificial Intelligence in healthcare has huge and wide-reaching potential. Learn more about how Oracle’s Smart Health Solutions can help by clicking here.
Enjoyed reading this blog? Leave feedback for us.