Finding new medical therapies for existing diseases is becoming more and more challenging. Safe drugs for many common illnesses have already been found. Now, scientists are looking for new ways of treating incurable conditions. But this search isn’t easy. Nine out of 10 drugs that are developed fail clinical trials in the last phases. High costs of research don’t make things easier. According to Nature, every dollar spent on medical research and development yields less than two cents of realized value.
For this reason, more and more scientists are pinning their hopes on smart algorithms. The use of AI and machine learning in healthcare puts a new perspective on the future of drug development. It’s already redefining the current medicine available.
It takes years of medical training to quickly and correctly diagnose diseases. Machine learning could speed up that process. It's already being used in diagnostic imaging to help identify patterns of tuberculosis or cancer. Using ML in pattern recognition is especially useful in developing countries where medical specialists have had little training in diagnostic imaging. Used on a wide scale, ML algorithms could help plenty of pathologists identify the onset of many serious diseases faster.
It’s worth mentioning that algorithms used in pattern recognition are only as good as the data they use. That’s why more and more tech companies are combining forces with pharmaceutical firms such as Bayer. By doing so, tech giants get access to huge amounts of medical information that is needed to improve the accuracy of their algorithms.
What’s more, machine learning is being used to fuel solutions such as the Face2Gene app. It’s facial recognition software that utilizes ML algorithms to identify phenotype, a set of characteristics typical for genetic conditions, in a patient’s photo. The research conducted by The Lancet Digital Health shows that well-trained algorithms can diagnose a patient with the same accuracy as medical specialists.
Monitoring an individual’s health is becoming much easier thanks to wearables. These are various types of electronic devices people wear, like smartwatches or wristbands, to collect information about their bodies. According to Statista, in the past three years, the demand for wearables has more than doubled, growing from 325 million in 2016 to 722 million in 2019. Nothing is surprising about that. Wearables are easy to use and provide real-time data about a person’s health status. These characteristics make them very popular among sports professionals and amateurs alike.
Wearables have turned out to be helpful tools for physicians too. They gather a huge amount of data about a person’s health that could help doctors better treat patients with chronic diseases.
A good example of wearable technology is Current Health. It's an AI wearable device that measures a patient’s pulse, oxygen saturation, temperature, respiration, and mobility. The tool uses machine learning to analyze collected data and identify anomalies. Current Health has already been approved by the US Food and Drug Administration agency and is suitable to be used at home. The device is especially helpful for patients with heart and pulmonary diseases, which are considered to be the most fatal in the US.
Thanks to wearables, patients can monitor their health in real-time. This allows for identifying warning signs much faster, contacting their doctors, and, in many cases, saving their lives. On top of that, the more doctors know about the state of the patients’ health, the more effective treatment they can recommend. This is beneficial for patients and for healthcare providers as well. More effective therapies reduce the cost of visits and hospital readmissions.
Biotech companies are using AI to give their drug development a leg up. With a huge amount of data being provided, machine learning algorithms can help in testing and discovering new medications for rare diseases. Deep Genomics, a Canadian biotech giant, has been using its AI platform to develop a treatment for Wilson disease. It’s a serious genetic disorder that causes the accumulation of toxic levels of copper in the body.
Based on machine learning, the Deep Genomics platform has identified, in a relatively short time, genetic mutations that cause the disease. The discovery has opened the door to new therapeutic opportunities for people affected by the disease.
Checking symptoms and managing medications
24/7 medical care is very expensive for hospitals and healthcare providers. With the well-being of their patients in mind, more and more health institutions are implementing chatbots. These smart assistants are a good way to support patients when access to doctors is limited. Moreover, when feeling unwell, many patients do research online. This often leads to misdiagnosis. Verified chatbots are a much better solution to triage patients and guide them to receive proper medical help.
A good example of this type of solution is LiveChat’s Covid-19 Risk Assessment Chatbot. It’s a free tool that helps patients pre-diagnose their coronavirus symptoms. The chatbot's content is based on the Risk Assessment tool built by Infermedica, a provider of AI-driven solutions for preliminary medical diagnoses. The solution helps companies and institutions share verified information about the COVID-19 symptoms and provides patients with contact information for local healthcare organizations.
Chatbots can also help patients maintain good health, like Florence, a Messenger chatbot. Florence helps patients remember to take their medicine and lets them track their health. It educates patients about their diseases and motivates them to stick to good habits.
80% of patients who are treated with the 20 most-used prescription drugs in the U.S don't respond to the medication. Doctors still know too little about many diseases and their patients which very often doesn’t allow for tailoring the method of treatment to the patient’s condition.
The problem has already inspired tech companies to look for ways to develop so-called precision medicine. The term refers to designing a targeted treatment for small groups of patients based on their characteristics, rather than the entire population.
An example of this type of study is Project Hanover, which was launched by Microsoft. It aims to help oncologists design effective and individualized cancer treatments. Microsoft researchers use machine learning algorithms to analyze a massive volume of cancer research. This makes it possible to determine how cancers develop. This accumulated knowledge could help scientists develop more personalized treatment plans and, eventually, heal more patients.
Predicting the outbreak of epidemics
AI can be of great value for organizations that track health epidemics as well. In 2014, HealthMap, an online mapping tool developed at the Boston Children's Hospital, detected an Ebola outbreak in Africa. It was nine days before WHO announced the pandemic.
HealthMap uses algorithms to search social media, news pages, and government sites to track instances of diseases on a map. Because of that, it can detect pandemics before epidemiologists are able to analyze massive amounts of data.
We need to keep in mind that tools like HealthMap don’t work flawlessly. It’s possible that they could distort results. Nevertheless, their ability to quickly gather and analyze data might help epidemiologists identify the spread of viruses much sooner in the future.
The medicine of tomorrow will be smart
AI and machine learning are making healthcare more efficient. They help identify diseases and design personalized treatments faster than ever before. It’s important for patients and medical communities to understand that AI isn’t meant to replace human specialists. Its objective is, above all, to support medical experts in making their work more effective. However, this will only happen if companies developing AI get access to enormous amounts of medical data. To become dependable algorithms, just like people, they need to learn a lot by analyzing real cases.
It’s time for health organizations and governments to take steps towards improving digital health literacy. The sooner we adopt AI and machine learning into healthcare, the more sustainable the system will become. Healthier, and happier, societies will come after.