AI Health Frontier – Issue 4

“Artificial intelligence is a tool, not a threat.”

– Rodney Brooks


Artificial intelligence in healthcare

DeepMind’s AI could effectively detect more than 50 eye diseases 
The computer was asked to give a diagnosis in the cases of 1,000 patients whose clinical outcomes were already known. The same scans were shown to eight clinicians. The AI solution performed as well as two of the world’s leading retina specialists, with an error rate of only 5.5%.
[BBC / Fergus Walsh]

A new pacemaker hack puts malware directly on the device
A chain of vulnerabilities in a certain type of pacemaker could potentially be exploited to control implanted pacemakers remotely, deliver shocks patients don’t need or withhold ones they do and cause real harm

[WIRED / Lily Hay Newman]

How new technologies could transform Africa’s healthcare system
While 65% of Africa’s working population still works in farming, AI technology might be of great help to the healthcare sector in Africa. If AI systems could handle some of the minor healthcare issues, the available healthcare professionals could focus on the more difficult issues.
[Harvard Business Review / Ndubuisi Ekekwe]

3 ways health AI is changing the medical field
According to a recent report by Accenture, AI-assisted surgery, virtual nursing assistants, and the management of administrative workflow are the top 3 most valuable AI applications now taking form in healthcare.
[EdgyLabs / Zayan Guedim]

Futurists in Ethiopia are betting on AI to drive development
While Ethiopia has been encouraging investments in the manufacturing sector, AI company iCog aims to place artificial intelligence at the heart of Ethiopia’s rapid development.
[Quartz / Thomas Lewton]

How VR could help people with dementia find their way around
For people with Alzheimer’s and other degenerative diseases, just navigating around the house can be difficult and disorientating. But some pioneering approaches using VR are offering new solutions.
[The Guardian / Jules Montague]

AI-driven dermatology could leave dark-skinned patients behind
Machine-learning algorithms could help diagnose skin cancers and other skin issues. But the data for the system comes from primarily fair-skinned populations in the United States, Australia and Europe. If the algorithm is basing most of its knowledge on how skin lesions appear on fair skin, then theoretically, lesions on patients of color are less likely to be diagnosed.
[The Atlantic / Angela Lashbrook]


Interesting stuff in the world of AI

Nutrigenomics – could you genes choose the right cheese for you?
Nutrigenomics looks at your genetic makeup and how it affects the way your body processes food. But to what extent do your genes really determine what food you should and should not eat?
[The Medical Futurist / Berci]

Meet Deeplocker, the AI weapon that can target specific individuals
To study how AI could be used as a weapon, IBM developed Deeplocker: malware that’s carried by systems like video conferencing software, which is dormant until it identifies its specific target through facial- and voice recognition.
[ZD Net / Charlie Osborne]