Dear friends,
Our three projects – Soil Quality Evaluation System, General Computer Vision for Healthcare, and CoVision – are running and show considerable progress! In addition, the number of volunteers at MI4People and our network of NPOs and fighters for the Public Good are constantly growing.
So, we look back on the work done in the last month with a lot of gratification. Nonetheless, one important component of our mission was missing till now – a broader public discussion about AI for Good movement, in general, and the work of MI4People, in particular. So, we were very happy when several weeks ago an AI influencer and podcaster, Luke Whipps, contacted us with an offer to produce an episode of his AI Game Changer Podcast about MI4People.
And this month the MI4People podcast is finally released! In this episode Luke discussed with our cofounder and managing director, Paul Springer, on topics like AI for Good movement, motivation of data scientists and Machine Intelligence (MI) experts to apply their skills to improve the world and our society, and of course about our work at MI4People, our philosophy, and our projects.
You can listen to this podcast on Spotify or watch it on a brand-new YouTube channel of AI Game Changers Podcast. In fact, MI4People had the honor to be on the first Luke’s podcast episode that was recorded not only as audio but also as video 😊
We hope you will enjoy this podcast and will engage in the public discussion about AI for Public Good!
In addition, and as we do every month, we have collected some interesting news about how Machine Intelligence can be used to foster delivery of Public Good. This time we focus on how MI can help fight one of the biggest still unsolved healthcare problems of our time – cancer.
So, happy reading this newsletter, put your capabilities to help Public Good delivery into action, and let us together make the world a better place for all of us!
Your MI4People Team
News
AI and Fight against Cancer
Just in 2020 alone, globally, there were ca. 19.3 million new cancer cases and almost 10.0 million cancer deaths. So, it is not surprising that, e.g., in US cancer is the most feared diagnosis. And this even in the time of the COVID-19 pandemic!
The good news is that AI can help fight cancer. The application of AI in this domain includes supporting doctors – especially unexperienced ones – in identifying right diagnoses, prediction of individualized cancer risks so that patients can take the most suitable preventive and screening measures and finding new and tailor-made treatments of cancer. In this newsletter we provide examples from each of these categories of cancer related medical activities.
Detecting Cancer with AI
Many cancer types are diagnosed by means of medical imagery such as x-ray, CT and MRI scans. Usually, you need an experienced professional to identify the cancer with high precision, especially in the early stages. Unfortunately, such experts are in short supply especially in poorer regions of this world.
Computer Vision, a subarea of Artificial Intelligence, that is already used to detect and classify objects on images and video, for face recognition on mobile phones, and in self-driving cars, is also able to identify diseases, including cancer, from medical images. The accuracy of such diagnoses is already at human-expert level and sometimes even can surpass an average human doctor (s., e.g., this paper on breast cancer detection)!
There are a lot of specialized Computer Vision algorithms and techniques that can handle various cancer types like lung, breast, brain, skin, and prostate cancer. For a review on existing techniques, we encourage the reader to look at this survey paper.
Early Detection and Predicting Cancer Risk
To cure cancer, it is extremely important to identify it as fast as possible. Unfortunately, many techniques that are used during cancer screening like x-rays are by itself harmful. Therefore, it is not a good idea to perform such screening too frequently. To resolve this dilemma, we must better understand and predict which patient is at high risk of getting cancer to be able to create screening strategies that achieve earlier detection with as less as possible screening harm.
An international researcher team has shown on example of breast cancer and mammography that Machine Learning (ML) can be used to predict risk of getting cancer. They created an AI model that uses a person’s mammogram images to predict their risk of developing breast cancer in the next 5 years. In various tests, the model was more accurate than the current tools used to predict breast cancer risk. Based on this risk estimation, a doctor can develop a strategy for how often someone should get screened for breast cancer.
Fostering Individualized Cancer Treatment with Machine Learning
While researchers and medical practitioners could considerably reduce the mortality rate for people with cancer over the past decades by means of novel treatments, it is still a challenge to identify which treatment is most promising for a particular patient. Therefore, medical practitioners often must try out several treatments on a patient before they can find the one that is impactful in his/her particular cancer situation. Beside high expenses, this costs a lot of time which is critical when treating cancer.
The answer to this problem could be personalized cancer treatment with help of Machine Learning (ML)! An innovative partnership between The University of Texas at Austin’s Machine Learning Lab, Oden Institute for Computational Engineering and Sciences, and Dell Medical School aims to speed up the process of identification of most suitable cancer treatments by means of combination of computational oncology and ML. While computational oncology uses advanced mathematical and computational simulations to model tumors, calibrate patient-specific models, and simulate patient responses to potential treatment options, Machine Learning algorithms can be applied to large data sets to build classifiers that can make accurate individual predictions, even in complex biological and chemical systems. Combining both technique is a very promising approach for developing personalized, individual-dependent treatment strategies, but also for finding completely new ways to treat cancer. For more information on this research initiative, read this article from UT Austin.
Concluding Remarks
Machine Intelligence can help to achieve the next breakthroughs in our fight against cancer. In addition, it can help overcome the shortage of specialized medical staff such as experienced oncologists and radiologists required to provide right diagnoses. Especially in developing countries that suffer from tremendous lack of medical experts, such MI diagnostics systems might save a lot of lives. In fact, this problem is so big that we at MI4People have created a dedicated project for this domain. The project is designed with a broader scope and aims to identify various diseases from medical imagery in developing countries and will also help combat cancer in these regions of the world.
Comments