With the power to do otherwise repetitive and monotonous aspects of the job, it hints at a future in which physicians can save more lives than ever.
Mendel.ai, named after the father of modern genetics, Gregor Mendel, is the brainchild of Dr Karim Galil. Tired of messy medical records, losing patients to cancer and trying to find which new clinical trial touts a solution that will cure which patient, Galil created the organising, artificially-intelligent system.
Mendal.ai skims through scientific papers to find the latest, cutting-edge medical research, using an algorithm that can understand the language naturally found in scientific papers. It then compares them with patient records to find the best match and assesses the patient’s eligibility for each suggested treatment.
Taking only a few minutes, the technology will be extremely beneficial for busy doctors. Galil predicts it will tremendously help patients as well.
“A lung cancer patient, for example, might find 500 potential trials on clinicaltrials.gov (American), each of which has a unique, exhaustive list of eligibility criteria that must be read and assessed,” Galil said.
“As this pool of trials changes each week, it is humanly impossible to keep track of all good matches.”
2. Spotting heart arrhythmia
A team of researchers from Stanford University, led by Andrew Ng, have designed a machine-learning model that uses an algorithm which can identify heart arrhythmias from an electrocardiogram (ECG) better than a doctor.
This type of machine is called a deep-learning machine and it functions when large quantities of data are fed into a big simulated neural network and its parameters fine-tuned until it can recognise unusual ECG signals. To develop the technology, the team partnered with a company called iRhythm that makes ECG devices.
They assessed the accuracy of their algorithm, by collecting 30,000 30-second clips of different forms of arrhythmia from patients and compared the machine’s ability to identify them with that of five different cardiologists. Then, a panel of three expert cardiologists judged who fared better.
Patients are usually asked to wear an ECG sensor for several weeks before a doctor can even begin to find an irregularity. Even then however, it can be difficult to distinguish between irregularities that may be life-threatening and those may not be.
It is hoped that the automated approach will make diagnosis more reliable and good quality care more easily available in areas with poor healthcare. For now however, Ng is “encouraged by how quickly people are accepting the idea that deep learning can diagnose at an accuracy superior to doctors in select verticals.”
CareSkore is another AI technology that uses machine-learning, this time - to anticipate mortality. By combining clinical, socio-economic, demographic, and behavioral data, it hopes to form a holistic picture of each patient to be used in preventative care.
The idea is that the AI will be able to inform doctors of patients who are more likely to skip appointments and forget or intentionally fail to take medication. Additionally, it is hoped that the technology will be able to predict the risk of a patient’s readmission to hospital and infection risk. Doctors can then use the information to reach out to patients with help and advice.
The technology will also be able to send patients personalised text messages and phone calls if anything is amiss. It will also be able to make suggestions for care of long-term conditions such as diabetes. The system can also be used the other way around so that patients can inform their doctors of any new symptoms and ask any questions they may have of their care.
The information collected will be used to further develop CareSkore’s predictive abilities and by using sources for environmental data such as Google maps it can put patient data into context. MIMS
Watson: Transforming healthcare with Artificial Intelligence
Can AI revolutionise drug pricing and supply the next generation of drug pipelines?
3 diagnostic mysteries solved through DNA sequencing