Computer vision delivers faster and more accurate patient diagnosis
These continued advances in AI and deep learning are revolutionising certain aspects of the healthcare industry. In particular, medical imaging is quickly evolving thanks to the use of computer vision which enables AI to identify patterns and extract key insights from data. Powered by deep learning and by leveraging AI algorithms that process images, computer vision is helping to pinpoint health conditions in medical images and is dramatically improving the time it takes to make a diagnosis with more accuracy than a physician.
Computer vision has several different applications in healthcare, particularly in radiology and oncology. It can be used to monitor tumour progression and rapidly identify metastases in tissues. Tools like IBM’s Watson Imagining Clinical Review use AI to aid physicians in medical image interpretation and thus increase accuracy while driving down costs.
Although the benefits of these technologies are well understood, adoption has been slow since incorporating these new solutions into existing infrastructure and workflows entails challenges. The healthcare system itself continues to be an underestimated hurdle. This means that insurers are still waiting for the technological breakthrough that will be successfully implemented into the healthcare infrastructure.
The use of AI and machine learning has the potential to completely transform clinical decision-making and bring significant changes in the diagnosis of health conditions and treatment. Tools have been developed that can process a large amount of patient data, make suggestions for the appropriate next steps in patient care and identify potential risks and issues (e.g. medication allergies) before they arise. Vendors such as VisualDx, offer system software for clinical decision support that enables physicians to make evidence-based decisions using AI tools.
VisualDx’s AI solution, DermExpert™, has proven to be as effective at diagnosing skin conditions in patients as primary care providers. Clinical decision support tools can not only augment physician decision-making, but also aid in reducing the occurrence of medical errors and ultimately improve patient outcomes. In theory, an application to help improve clinical analysis is good news for doctors, patients and insurers. One practical barrier to overcome is that the quality of data that is fed into the algorithm is still not sufficient to derive valid treatment choices considering the complex reality of a patient’s medical condition. What insurers want to know: How can data quality for machine learning be improved in order to support the diagnosis process?
Sequencing of human genomes and personalised medicine
With the cost of genomic sequencing decreasing dramatically and the technology behind it improving significantly in recent years, the availability of personalised medicine is becoming more mainstream. With the introduction of automated high-throughput sequencers, AI speeds up the processing of huge genome datasets and helps in identifying potential genetic disorders.
This new accessibility can be seen in companies like 23andMe or GenePlanet, which offer consumers genetic testing concerning an individual’s risk for certain diseases. Additionally, companies such as Oncompass offer patients more bespoke treatments by giving biopsy analyses based on their genetic profiles.
While this technological advancement offers more transparency to patients and customers, and a clearer outlook on prospective future health risks, regulations often forbid the use of this data for life or health insurance – especially in Europe. An insurer could face the risk of anti-selection in the future, when more and more people utilise genetic testing before signing up for an insurance policy. There is a possibility that regulations will eventually change and allow the use of genetics in insurance contracts. If this happens, however, the information will have to be considered and incorporated into the insurance pricing.
Artificial Intelligence and related developments will continue to help revolutionise healthcare. The number of existing use cases in the medical field is enormous and ranges from suggesting an optimal treatment plan to reducing diagnostic errors. But recent cases also show limitations and challenges which must be overcome before they can become relevant for a huge number of patients. Insurers need to stay aware of these leaps in healthcare technology while observing how they are transforming the insurance business – perhaps slowly, but surely.