As one may easily guess, a ‘Medical Imaging AI’ company makes revenue from medical imaging, especially diagnostic tools. However, this conclusion should not be drawn too early. In this era, data is money.
Many companies in the world have been founded based on business models that they might no longer practice later. Take an example: 3M is a company famous for worker safety, health care and consumer goods. However, it was established initially as the Minnesota Mining Manufacturing Company, a mining firm.
Artificial intelligence (AI) is a topic that is frequently brought up in this decade, from university settings to corporate boardrooms. In the healthcare industry, AI has attracted tons of cash from venture capital and corporate funds. As AI requires massive structural data, medical imaging has become a field that draws attention from AI startups. Medical images come from X-rays, CT scanning, MRIs, etc. With previous radiographers’ notations and classification, medical imaging has accumulated massive data that can be used by AI companies.
2019 being different from 2018’s investment frenzy, AI medical imaging companies experienced a peaceful year: no dramatic financing nor waves of bad news about dying companies. In 2019, the financing that was injected into AI medical imaging companies dropped to just over 1/3 of the number of 2018. It is interesting that most financing activities are at an early stage – Series A financings taking the biggest chunk.
As AI’s heat becomes ‘gone with the wind,’ AI companies are having to stand on their own feet to survive the winter. Considering that healthcare is a special industry, regulation is considerably stricter than other industries. Patient privacy, disease diagnosis, health management and other aspects are all sensitive matters playing on citizens’ nerves. The regulatory system aims to ease the public’s related worries and build a reliable mechanism to assure the whole industry runs without fault.
Governments are exploring the fine line around regulating AI’s function in healthcare. Without a specific class, national healthcare insurance and commercial insurance will not be able to price the service – state-owned hospitals, meanwhile, cannot employ these services in their daily operation, due to regulatory reasons. The CFDA classified AI medical imaging diagnosis assistance as ‘Class III software’ under the Classification Catalogue for Medical Devices (Classification) in 2017. The new classification seems to point a way for AI startups to realize revenue. However, the market has not demonstrated a favorable attitude.
Under Classification, AI medical imaging analysis software can only assist radiographers in diagnosing. AI-based decision-making is still a grey area from the perspective of legislation, as the law requires a specific subject to be reliable for the act. The AI assistive diagnosis service helps with radiographers’ efficiency in speeding up diagnosis but cannot help to soothe the tension that comes from imbalanced healthcare resource distribution nationwide. Besides, filing for CFDA’s medical device classification can take years. Before acquiring the license, how do AI medical imaging companies make a living?
The question leads to a side revenue source for these companies – radiographical-related research. Modern science is built on experiments and data analysis, and AI makes data speak louder. With the assisted diagnosis provided by these AI startups, medical research can assess more data and save labor costs on sorting and analyzing data. For instance, from RSNA site’s 12Sigma (图玛深维) and Infervision(推想科技) intro, cooperation with research institutes, including academic organizations and clinical research hospitals, is another way to make contributions to radiography while diversifying income structure. According to 12Sigma’s homepage, it has cooperated with over 30 clinical hospitals for research purposes.
Mature medical tech companies, like Siemens Healthineers, GE Healthcare and Phillips Healthcare, have been actively developing AI technologies and applying it to their product lines. For these giants, AI medical imaging reading not only provides a better experience to their clients in the diagnosis process, but also can use the tech as feedback to improve existing products like MRI and CT scan machines to generate a better image with less noise and improved accuracy.
Pros and cons are the two sides of the coin. What concerns the public the most for AI and other technologies related to big data is privacy protection. Bloomberg reported earlier that Silicon Valley is involved in a privacy crisis when improving smart voice assistants’ capabilities. It is commonly known that AI is based on big data, and to improve the intelligence level of AI, there is a growing demand for real data, which might put pressure on people’s inner peace – who will feel safe if their data is permanently streaming to unknown servers? But without being fed by real data. AI will not evolve as intelligently as expected. So, we face a choice, between making: advanced AI that makes our lives more comfortable or a world in which we keep our data to ourselves.