Healthcare is one market where end users can have their needs and requirements addressed with advanced AI solutions.
Needless to say,
AI can help end users solve various security issues that they are faced with. Yet beyond security, AI also has a wide range of use cases across different verticals. In particular,
healthcare is one market where end users can have their needs and requirements addressed with advanced AI solutions.
One of the major issues facing healthcare providers is safety. This is especially the case after a rise in patient falls in various healthcare facilities, senior homes and other assisted living organizations. Amid this problem, how to detect patient falls early and respond in a timely manner has become important.
In this regard, AI can play an important role working in conjunction with various technologies.
Video surveillance with pose estimation analytics, for example, can alert operators of unusual pose positions, such as falls, suffered by patients.
Yet there are certain difficulties or limitations with these technologies. Video for example can be a source of concerns especially in facilities with heavy
privacy requirements. Contact devices such as smart watches or bracelets worn by patients meanwhile have power and transmission issues.
In this regard, Taiwan-based IOEZ stands out by providing AI in conjunction with millimeter wave (mmWave) and time-of-flight technologies in healthcare applications.
“For leave-bed and vital sign detection, we use mmWave to measure the speed and angular characteristics of objects, and develop accurate AI algorithms. For the fall detection, we use ToF to measure the speed and distance of objects and the characteristics of point cloud imaging to develop accurate AI algorithms,” said Alan Li, Director of IOEZ. The company’s main targets are medical facilities, long-term care centers and day care centers, especially those in Taiwan and Japan which have a large population of elderly people. It pushes its technology through SIs in the medical industry.
While the company claims that “we almost have no competitor, as no other company combines AI algorithms with mmWave radar and ToF sensing technology,” there are still certain challenges that the IOEZ is faced with. One of the biggest challenges is the sensor equipment’s high cost, which affects the end user’s purchasing intention.
According to Li, the company overcomes this challenge in two ways. “First, we provide the systems for medical institutions to test. We hope that by showing actual benefits to push hospitals will be willing to buy. Second, we maintain a very good relationship with the sensor providers,” he said.
Abnormality detection
Beyond safety use cases, another main application of AI in healthcare is detection of abnormalities in human bodies. Specifically, several types of cancer, such as breast cancer, have better prognosis if detected early. Yet standard screening procedures, for example mammograms, can still produce false negatives – meaning the screening does not detect the cancer when the patient actually has it. With AI, accuracy can be significantly improved. According to a study by Lunit, AI detected 91 percent of T1 cancers and 87 percent of node-negative cancers, whereas the radiologist reader group detected 74 percent for both.
One company that specializes in this area is Taiwan-based AIExplorer, whose AI is being used to detect various cancers including breast cancer and lung cancer. It is also used in other medical applications such as fetal ultrasound, diabetic retinopathy, dental X-ray image analysis and sleep apnoea diagnosis.
Customer request was cited as the chief rationale for AI development by the company, which offers both server AI solution as private cloud AI and edge AI solution as in AI boxes. While similar solutions exist, the company excels with speed and accuracy. Specifically, the company claims to be the world’s leading terapixel platform, with which AI models could be improved continuously by adaptive learning in minutes – 20,160 times faster than existing AI systems, which tend to cost two weeks to train an AI model on terapixel images.
“We provide high-quality AI with superior performance on speed, accuracy, stability and reliability of the system,” said Ching-Wei Wang, Founder of AIExplore, adding that besides medical use cases, AIExplore also targets other applications. “These include concealed weapon detection, license plate recognition by both portable and fixed cameras, human detection and face detection in long distance, computer vision measurement, defect detection and suspicious behavior detection,” Wang said.