With the cancer vaccine scheduled to be tested in humans at the end of this year, and new AI-driven advanced detection techniques, we're getting closer than ever to winning the war against cancer. We can now predict this most dreaded disease before it occurs, and treat it with new drugs that can target the unique DNA weaknesses of that specific malignancy.

Early Detection

Spotting cancer as early as possible is of paramount importance. If a tumor is diagnosed at an early stage, doctors can treat it with a much higher chance of success before it gets too big. The more a malignancy has spread, the lower the patient's chances of surviving. In a previous article, we already talked about algorithm-based software that can analyze every kind of medical imaging report to spot even the most minuscule anomaly that the human eye can't hope to find. Some of them are so exact that they boast an amazing 88 percent detection rate, and can be used retroactively to check all the previous medical records of a given patient (or even a population) in a matter of minutes.

Newer intelligent algorithms that can spot complex tumor patterns are being developed every day, and some of them can be used to detect a tumor as early as the very moment in which it is formed. A cancer therapy startup called Cyrcadia Health developed small, wearable patches that could be comfortably inserted under a bra to detect temperature changes within a woman's breast. Using machine learning predictive analytics software, the smart device can detect any abnormal circadian patterns in breast tissues and immediately alert the woman (and her healthcare provider). According to the early tests made by the manufacturer, the sensor-filled patches can detect up to 80 percent of breast tumors. (For more on how tech is being used in health, check out The Role of IT in Medical Diagnosis.)

What's even more interesting, is that machine learning is bound to open new opportunities for early detection in due time. What makes cancer a disease which is so hard to deal with is the extreme variability of its many forms. Although many great advances in cancer genomics have been made, monitoring the human DNA to spot any genome mutation requires substantial efforts in sequencing. The more malignancy samples and examples AI can collect, the more it can learn about cancer and significantly ease the computational burden of sequencing any potential mutation.

Improving Existing Treatment

Most traditional chemotherapy agents are known for their devastating effects on the human body, such as alopecia, constant fatigue, pernicious vomiting and many others. Newer, more selective biological therapies have been engineered in the last few years, to stimulate the body's immune system to act against malignant cells. Referred to collectively as “immunotherapy,” many of those newer therapies are much more tolerable, but it is hard to predict whether they will work against a specific tumor or not.

One such example is PD-1 inhibitors, a group of monoclonal antibodies that act by preventing cancer cells from deactivating the immune system. However, some patient populations are known for their extremely low response rate to this type of treatment. For example, PD-1 inhibitors do not work in about 80 percent of advanced non-small-cell lung cancer patients, leading to a significant waste of resources due to the high cost of these antibodies.

Precision oncology is a new branch that develops new techniques that improve treatment decisions by finding, for example, only those patients that could benefit from the above-described treatment with PD-1 inhibitors. Researchers at Institut Curie in France are working with the American startup Freenome to develop a new non-invasive alternative to surgical biopsy to scour for cancer DNA circulating in the blood. Freenome's AI is fed with data coming from cancer patients and is tasked with the goal of finding any link between blood biomarkers and the patient's response to treatment. Their clinical trial could be the first among many aimed at improving the efficiency and precision of modern immunotherapy, saving precious resources that are wasted treating patients who would not benefit. (Tech is becoming more prevalent in health care, but what do patients think of it? Check out What Do Patients Want From Health Care Technology?)

Finding New Cures

The so-called "cancer vaccine" that, so far, has cured up to 97 percent of tumors in mice, is probably the most groundbreaking news in ages. Actually just a much more precise form of the above-described immunotherapy, the cancer vaccine gets its name from the fact that it can prevent tumors from coming back. Once again, this new amazing treatment actually activates the immune system's T-cells to eliminate cancer cells throughout the body. What makes this new "vaccine" different from other types of immunotherapy is that the two agents that compose it are injected directly inside the tumor to reactivate the "dormant" T-cells. Because of that, these cells are not like any other T-cell found inside the body, but a specific population that has been trained to recognize cancer-specific proteins. Once they destroy the tumor inside that tissue, they can even roam freely via blood circulation to search for and destroy any other cancer cell that has infiltrated other tissues (a phenomenon known in medicine as "metastasis").

If this idea sounds incredible, well, that's because it is. Are we going to win the war against cancer as soon as this vaccine completes its trials and is released to the public? Sadly, things are seldom so simple, and this treatment only works on a specific subset of cancer types, because each type of cancer is affected by the immune system in a different way. And that's where AI is going to help us, once again, as a deus ex machina, or, in this case, a machine-learning deus ex machina.

The Danish company Evaxion has been recently granted an almost $1 million fund to develop a machine-learning project that would allow immunotherapy to be customized the individual patient's needs. The mutations that lead to the uncontrolled growth of malignant cells differ from patient to patient, and depend on his or her specific genome. By sequencing genes in cancer cells and healthy cells from the patient, the AI can identify the individual DNA changes specific to that patient's cancer, and then design vaccine antigens that, once again, lend a precious hand to the host's immune system.

Evaxion is far from being the only company looking for customized solutions in cancer therapy, and the only thing that really differentiates the various startups is not the method, but the potency of their machine-learning algorithms. Whether it will be the Danish company that would eventually win the race, only time will tell, but what really matters is that the element that is going to make the difference is the AI.

Conclusion

One of the highest, most insurmountable walls that currently makes cancer therapy a privilege available only in the richest countries or to the wealthiest individuals is, by far, its exorbitant cost. These new AI-powered technologies can help reduce the waste, and have the potential of dropping the costs, making cancer treatment much more affordable and, in turn, more "democratic."