Tue, Jan 30, 2018 - Page 8 News List

AI is here again, this time to stay

By Jack Li 李友專

Humanity is aspiring to transcend old age and death. In the 21st century, this might no longer be an unattainable dream.

With everyday exposure to news media, movies, novels and theater, it is easy to assume that artificial intelligence (AI) will be able to provide all-powerful machinery capable of solving matters such as birth, aging, sickness or death — but how far has AI actually progressed in terms of medical science?

As top medical journals show, the progress made by AI goes no further than the ability to more meticulously analyze retinal changes as a result of diabetes in angiographic images; more precisely identify melanoma in dermoscopy images; find lymph node metastasis through analyzing histopathology images of breast cancer patients more accurately and faster than pathologists; forecast blood sugar levels by making calculations based on human genes and gastrointestinal microbiota; and assist oncologists’ decisionmaking by providing more informed and concise suggestions for cancer treatment, an AI technology provided by IBM’s Watson for Oncology.

The IBM Watson after its launch in 2016 received polarized opinions following its failure at the University of Texas’ MD Anderson Cancer Center, with some considering its AI capacity to be immature.

However, the higher the expectations, the greater the disappointment when the technology fails.

Just like the aforementioned advanced medical applications for AI technology, the IBM Watson for Oncology is by no means all-encompassing. It was not designed to diagnose cancers, but to target seven kinds of cancer and provide more in-depth and specific treatment advice. It also cannot give advice on surgery and radiation therapy, nor can it offer suggestions to patients with cancer recurrence or metastasis.

It might sound as if the IBM Watson is useless and not nearly as miraculous as the public believes it to be, but when it comes to treating people diagnosed with early stages of one of the seven targeted kinds of cancer, it provides detailed suggestions for medication and treatment, including chemotherapy, targeted therapy and immunotherapy.

Its suggestions are not only in accordance with the therapy already under way, but it also offers new directions and, in some cases, therapeutic opinions.

Within the scope of its database, IBM Watson is knowledgeable and it keeps “studying” new findings from journals to enhance the precision and subtlety of its analyses.

If the question is whether AI technology can replace an outstanding oncologist, the answer is “of course not.” However, if the question is changed to “Will AI under certain circumstances be able to assist oncologists by providing more precise therapeutic opinions for cancer patients?” then the answer is probably “yes.”

Every time a new technology emerges, public expectations rise too high on the assumption that it will solve all current difficulties. In particular, people find it hard to resist imaginative temptation when it comes to a sexy phrase such as “artificial intelligence.” These wild fantasies often lead to what is known as “technology hype.”

In terms of AI applications in medicine, moderate technology hype could encourage entrepreneurs to make venture investments and bring new capital into the field, thus encouraging the world’s top 10 companies to put a large amount of funds and talent into research. After all, finding the solution to issues related to birth, aging, sickness and death is humankind’s Holy Grail and a tremendously huge business that is growing 10 percent annually.

This story has been viewed 3078 times.

Comments will be moderated. Keep comments relevant to the article. Remarks containing abusive and obscene language, personal attacks of any kind or promotion will be removed and the user banned. Final decision will be at the discretion of the Taipei Times.

TOP top