Author:
CEO & Co-Founder
Reading time:
Machine learning in healthcare is one of our favorite subjects. That’s why it repeatedly appears on our blog. The fact is, machine learning in healthcare is immensely important–it literally saves lives! Every year, we see more and more fascinating machine learning applications in the healthcare industry. Let’s examine some of the most interesting ones, and think about the shift that is fueled by the AI and ML technology.
There are dozens of incredible AI applications in the healthcare industry. Some of them have already been analyzed on our blog. Today, however, we want to analyze this question in a bit broader sense and see some of the real-life applications of this technology in service of our health.
We have a whole list of existing solutions and software, designed to protect our health and sometimes even lives. Let’s see what good does machine learning make in the healthcare sector.
Nowadays, we can indicate at least seven crucial fields in healthcare that can be significantly improved by the adoption of machine learning technology. In general, the goal is quite straightforward–to speed up the pace of physicians’ work and make it more accurate.
Today, many countries struggle with the overloaded healthcare system and a lack of experienced physicians. Machine learning comes to the rescue, although this technology can’t solve this issue altogether. For several years now, scientists have been working on machine learning models that predict disease susceptibility or aid in early diagnosis.
For instance, in mid-2019 it was revealed that MIT’s Computer Science and Artificial Intelligence Lab has developed a new deep learning-based prediction model that can forecast the development of breast cancer up to five years in advance[1]! Advanced prediction models can predict (based on historical data) what kind of diseases and illnesses a given patient is susceptible to.
Another company working in the same field is Quantitative Insights. They want to improve the speed and accuracy of a breast cancer diagnosis with its computer-assisted workstation called Quantx[2]. It’s the very first FDA-cleared software designed to aid in breast cancer diagnosis. It aims to reduce missed cancers as well as false positives that can lead to unnecessary biopsies.
Moreover, with ML, it’s much easier to diagnose each patient. KenSci is an AI company based in Seattle, US. Their systems use machine learning to predict illness and treatment to help physicians’ intervene earlier, predict population health risk by identifying patterns, and surfacing high-risk markers, and model disease progression. Furthermore, their solution predicts future risks for optimal care![3]
Back in 2019, we wrote an entire article about this machine learning application in healthcare. Did you know that medical images are the largest data source in the healthcare industry? IBM estimates even that medical images account for at least 90% of all medical data![4] The problem is that someone has to take these pictures, store them, and analyze in order to detect cancer cells and other adverse changes in the healthy tissues.
And this is where AI comes into play. You see, after initial training (you have to teach the ML algorithm which tissue is healthy and which is attacked by the disease), machine learning algorithms can detect attacked tissues at an unreachable for humans pace. ML algorithms can interpret imaging data like an experienced physician. They can identify suspicious spots on the skin, lesions, tumors, and brain bleeds.
They are never tired, they don’t go on a medical, they can work 24/7, all year round. That’s why ML-powered medical imaging is an enormous step in healthcare. With this solution, physicians’ don’t have to devote hundreds of hours to analyze medical images. The algorithms do it for them.
At this point, you might ask why is that not a standard solution worldwide? We encourage you to read our article about medical imaging to find that out. Here, we can only say that healthcare data is still highly heterogeneous and incomplete. Training a good machine learning model with such variegate datasets is quite a challenge.
PathAI is a company based in Massachusetts, US. Their technology employs machine learning to help pathologists make quicker and more accurate diagnoses as well as identify patients that might benefit from new types of treatments or therapies.
Stanford University has made create a diagnosis algorithm for skin cancer. They made a database of nearly 130,000 skin disease images and trained their algorithm to diagnose potential cancer. Their algorithm matched the performance of the dermatologists[5].
Read more about medical imaging.
In the medical imaging section, we mentioned the healthcare data quality issue. It’s a severe problem that holds back the development of this field. To overcome this obstacle, we need to create a unified, high-quality patient database. Only then, it would be possible to devise a machine learning algorithm that’s capable of analyzing patient data on a global scale, and, thus, produce more accurate results and diagnoses.
What we need, is an AI-enabled patient data platform that’s able to connect to a multitude of various patient databases and to analyze patient data. It would be the next milestone in the healthcare industry. Unfortunately, it’s still yet to come.
Quotient Health, located in Denver, Colorado, is one of many companies working on ML-aided patient database. Using deep learning, Quotient Health creates a system that can optimize and standardize the way EMRs (Electronic Medical Records) are designed.
In order to reach truly personalized medicine, you need vast amounts of data about each patient. We are slowly moving forwards in this direction. Thanks to healthcare electronic data records, collecting patient data today is much more effective. With that data, it’s significantly easier to find the best treatment for a given patient. Machine learning (boosted by big data) helps in collecting demographic and medical data in reference to each patient, such as:
What’s more, machine learning can help in the prediction of disease incidences or detecting trends that lead to better health and lifestyle of society.
IBM Watson Oncology helps clinicians as they consider individualized potential cancer treatment options for their patients. IBM’s solution assesses information from each patient’s medical record and displays potential treatment options ranked by level of confidence[6].
Pfizer is one of the companies cooperating with IBM. With the help of IBM’s Watson AI technology, Pfizer uses machine learning for immuno-oncology research about how the body’s immune system can fight cancer[7].
This is another field thoroughly discussed in another blog post. We recommend you read it! As you know from that article, the first step in the drug development process is to understand the biological origin of a given disease. That requires a huge amount of data that needs to be processed.
With machine learning algorithms it’s a snap! They can analyze thousands of sources just in seconds. Today, machine learning algorithms are capable of designing aspirin and non-steroidal anti-inflammatory drugs, all on their own!
The truth is, ML can be applied at every stage of the new drug discovery process, from designing the chemical/protein structure to clinical trials. One of the key factors to conduct successful clinical tests is to find suitable candidates. Machine learning applications can considerably speed up the process of looking for these candidates and rejecting unsuitable ones. Also, machine learning drug discovery algorithms can also reduce data errors. The result? Faster, cheaper, and more accurate clinical tests.
Find out more about Artificial Intelligence in Drug Discovery with Machine Learning
Today, it is estimated that designing a single new drug takes 10-15 years and at least 350 million USD. No wonder that scientists all over the world are tirelessly working on designing applications and algorithms that can make this process considerably cheaper and quicker.
One of many excellent machine learning applications are the predictive models. By taking historical and present data into account, they can predict the chances of a given event or outcome. This finds application in healthcare as well. Today, many healthcare providers and AI companies are working on digital solutions that use anomaly detection algorithms to predict specific health events such as strokes, heart attacks, or sepsis.
These algorithms take into account historical patient data and real-time measurement of specific parameters, especially blood pressure, body temperature, heart rate, and respiration rate. If the risk of a specific event rises, physicians are instantly informed which allows them to take the necessary action as quickly as possible.
In El Camino Hospital in San Francisco, USA, researchers combined electronic health records, nurse call data, and bed alarm data to develop a tool for predicting patient falls. This tool alerts staff when the patient is at high risk of falling so that they can immediately take action. El Camino has achieved a 39% reduction in falls in just six months![8]
Last but not least, this application of machine learning in healthcare is particularly important today, in a world altered by the outbreak of the COVID-19 disease. Thanks to big data and machine learning, scientists can now assess the chances of malaria and other severe infectious diseases outbreaks. It happens by analyzing massive amounts of data, collected i.a. from:
BlueDot is a Canadian company that works on, what they call, outbreak risk software. Their algorithms have spotted what eventually came to be known as COVID-19, nine days before the World Health Organization released its statement alerting people to the emergence of a novel coronavirus!
Shortly after midnight on Dec. 30, 2019, BlueDot algorithms picked up on a cluster of “unusual pneumonia” cases happening around a market in Wuhan, China[9]. Unfortunately, we all know the results. But the truth is, this algorithm has worked 100% accurately. Today, the Government of Canada is using this tool for various purposes.
In this blog post, we mentioned a couple of interesting articles about machine learning in healthcare. We recommend you read them, to gain more knowledge on how machine learning is used in healthcare, especially drug development and medical imaging.
If you’d like to find out more, and maybe adopt this technology to your healthcare business, we’re at your service! Addepto’s professionals are always eager to guide you through this fascinating world and show you all the benefits that await you, just around the corner.
See our machine learning consulting services to find out more.
[1] Darrell Etherington. MIT AI tool can predict breast cancer up to 5 years early, works equally well for white and black patients. June 26, 2019. URL: https://techcrunch.com/2019/06/26/mit-ai-tool-can-predict-breast-cancer-up-to-5-years-early-works-equally-well-for-white-and-black-patients/. Accessed Jun 4, 2020.
[2] Anna Madrzyk. Artificial intelligence software for breast cancer diagnosis makes TIME’s list of Best Inventions for 2019. Dec 3, 2019. URL: https://www.uchicagomedicine.org/forefront/cancer-articles/artificial-intelligence-software-for-breast-cancer-diagnosis-time-best-inventions-2019. Accessed Jun 4, 2020.
[3] URL: https://www.kensci.com/solutions/care. Accessed Jun 4, 2020.
[4] Heather Landi. IBM Unveils Watson-Powered Imaging Solutions at RSNA. Dec 1, 2016. URL: https://www.hcinnovationgroup.com/population-health-management/news/13027814/ibm-unveils-watsonpowered-imaging-solutions-at-rsna. Accessed June 4, 2020.
[5] Taylor Kubota. Deep learning algorithm does as well as dermatologists in identifying skin cancer. Jan 25, 2017. URL: https://news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/. Accessed Jun 4, 2020.
[6] IBM. Supporting cancer research and treatment. URL: https://www.ibm.com/watson-health/solutions/cancer-research-treatment. Accessed Jun 4, 2020.
[7] IBM. URL: https://www.ibm.com/blogs/watson-health/. Accessed Jun 4, 2020.
[8] Mike Miliard. How predictive analytics, telehealth helped one hospital make patients safer. May 16, 2017. URL: https://www.healthcareitnews.com/news/how-predictive-analytics-telehealth-helped-one-hospital-make-patients-safer. Accessed Jun 4, 2020.
[9] Cory Stieg. How this Canadian start-up spotted coronavirus before everyone else knew about it. Mar 3, 2020. URL: https://www.cnbc.com/2020/03/03/bluedot-used-artificial-intelligence-to-predict-coronavirus-spread.html. Accessed Jun 4, 2020.
Category: