Dr. Joris Galland is a specialist in internal medicine and passionate about new technologies. After practicing at the Lariboisière Hospital, he joined the Bourg-en-Bresse Hospital (France). He explains the trends in medicine that unfold at present and which will impact our lives in the decades ahead. Originally published by our partners at esanum.fr
Cancer is the second leading cause of death in the world, with 8.8 million deaths in 2015, making it responsible for nearly one out of every 6.1 deaths. In France, the total number of new cancer cases has been increasing every year for thirty years. This is due to longer life expectancy (which has been correlated with a surge in the breast and prostate cases), behavioural and environmental factors, as well as improved diagnostic methods.
For some, cancer embodies a failure of modern medicine. However, medicine has made significant progress in cancer screening (PSA tests), diagnosis and targeted treatment of neoplastic cells. In half a century, scientific research has succeeded in significantly improving the prognosis of certain cancers with the arrival of biotherapies and gene therapies. Without forgetting the major role of prevention and/or screening campaigns, Artificial Intelligence (AI) is expected to revolutionise oncology in the next ten years. Here is how.
Early cancer detection is an essential element in improving a prognosis. The speed of diagnosis in imaging is already revolutionized by AI. While it takes about ten years of training to a human brain to learn radiology, an AI from Google manages to dethrone the best radiologists in the diagnosis of bronchial cancer after only a few days of training.2
Finally, what could be easier for an AI? The computer can image pixel by pixel, the human eye and brain are incapable of doing so. Thanks to its analytical accuracy and deep learning, AI becomes an ally of the radiologist. In 2019, a team of researchers published in the journal Nature an algorithm for predicting lung cancer risk from scanner images. When no previous image was available, AI did better than the radiologist with a 5% reduction in false negatives.
Wearables are another innovation, at the crossroads of the Internet of Things and AI. For example, we already have innovative solutions such as a bra that detects breast cancer. The American laboratory Cyrcadia proposes a prototype named iTbra™. This brassiere vest allows the detection of breast cancer in its early stages with at least as much sensitivity as a mammogram.3
Composed of two intelligent breast patches placed in a bra, this wearable technology identifies temperature changes in breast tissue. A self-learning predictive analysis AI is used to identify and classify abnormalities in breast tissue. In case of abnormalities, the system will alert the patient to go to a physician to perform the appropriate imaging examinations. The initial results of this invention are promising and certification is underway.
If this system is so efficient, why not use the same technology with a connected undergarment that would detect genital cancers? Or a T-shirt that would detect the presence of tumors? It sounds like science fiction, but the field of wearable technology is making great strides. The Hexoskin4 tee-shirt already allows the evaluation of a patient’s vital parameters, respiratory volumes, quality of sleep, etc. The device has been tested in COVID-19 patients, in the context of home rehabilitation or in research work.
The hope of oncologists does not rest solely on a very early diagnosis of the first cancer cells. Ideally, they would like to improve the prediction of cancer, even before the first cancer cell is formed. This is where advances in genetics come in.
In the early 1990s, leading researchers claimed that the human genome would never be sequenced, or at least not for hundreds of years. Advances in computer technology have made this feat possible: some machines can sequence the entire human genome in as little as four hours and for only a few hundred dollars (this is called Next Generation Sequencing, or NGS). Over the next ten years, sequencing speed is expected to improve exponentially and at modest cost.
If genome sequencing is no longer difficult, the interpretation of the masses of data generated becomes problematic. No geneticist is capable of interpreting this flow of information. This is where AI comes into play. A large genomic study carried out on nearly 10,000 women with ovarian cancer identified a common genetic variant increasing the susceptibility of tumor occurrence by 20% to 40%.5 With NGS, it would be possible to detect this variant in women, from birth, and offer them close monitoring if necessary. We are no longer far from the world depicted in the film Gattaca; and amidst the positive prospects, a legal framework becomes indispensable.
AI allows, among other things, the analysis of big data. It has the capacity to make links between a risk factor and a cancer, even though the human brain would not have had the capacity to establish a certain linkage on its own. For example, an AI has highlighted the link between the drug pioglitazone (a diabetes drug) and bladder cancer, leading to the withdrawal of the drug. If this algorithm, which takes only a few minutes to train, were deployed on a large scale, it would enable a real revolution in pharmacovigilance with the detection of side effects in near-real time.
One of the roles of the oncologist is to predict and diagnose the risk of relapse of a cancer in remission: here again, new technologies will be of great help. The MOOVCARE application allows to detect a potential lung cancer relapse.6 It analyses, on a weekly basis, the evolution of symptoms using a questionnaire completed by the patient.
This application uses an algorithm that achieves a relapse detection sensitivity close to 100%. Above all, relapses can be detected five to six weeks earlier than scans, which are typically performed every three months. The treating oncologist is then directly alerted. Tests are being studied for breast, kidney, prostate and lymphoma cancer. Thanks to its effectiveness and the medical service provided, the application is validated by the French main state health buro (Haute Autorité de santé or HAS) and covered by social security.
This is not an exhaustive list of connected health projects in oncology. Many start-ups are constantly working to improve the prediction, early detection, diagnosis and treatment of cancer. For these companies, the key element is the collection of big data. This is not a limiting factor for firms like the American GAFAMI or the Chinese BATX. These giants are therefore interfering more and more in the field of oncology.
IBM is starting to diagnose rare cancers and identify the best treatment thanks to its super AI Watson. For its part, Google X's biotechnology laboratory, named "CALICO", is arguably working on a project on ageing and health via the famous "DNA scissors" CRISPR-Cas9. The idea is to modify human DNA to abolish sickness altogether for the human species. Once again, the implementation of a legal framework becomes urgent, if we do not want to sink into eugenics.
Google is also working on a nanoparticle that could be diffused in the bloodstream in order, for example, to discover cancer cells or to fight them. This nanoparticle could even communicate directly with a person's watch to alert them if a disease is discovered.7
Tencent Holdings, the Chinese multinational company that operates, among other things, the messaging application WeChat, wanted to position itself in the digital medicine field. In 2017, it launched the Artificial Intelligence Medical Innovation System (AIMIS), a medical imaging system based on artificial intelligence. AIMIS enables the screening of several diseases, including diabetic retinopathy and certain cancers. The system, which is currently undergoing clinical validation in about 100 hospitals in southern China, has enabled physicians to analyse more than 100 million images. According to the firm, the image recognition accuracy rate is 90% for oesophageal cancer and 97.2% for colorectal cancer. The striking power of these giants, although unfair to small companies, will undoubtedly lead to spectacular advances in oncology.
References:
1. WHO - Cancer: main facts
2. Ardila D, Kiraly AP, Bharadwaj S, Choi B, Reicher JJ, Peng L, et al. End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med. 2019;25(6):954-61.
3. An introduction to the Cyrcadia Breast Monitor: A wearable breast health monitoring device
4. Hexoskin. Biometric garments for sports, research and health
5. Ford D, Easton DF, Stratton M, Narod S, Goldgar D, Devilee P, et al. Genetic Heterogeneity and Penetrance Analysis of the BRCA1 and BRCA2 Genes in Breast Cancer Families. Am J Hum Genet. 1998;62(3):676–89.
6. Moovcare®
7. Slate Magazine. Oremus W. Google Wants to Monitor Your Body With Nanoparticles.