AI Biweekly: Voice Assistants Take New Roles; AI Reduces Medical Overtreatment

Biweekly Review

Shan Ying – Analyst, Global Industry

Alexa into India, Cortana out of Competition

Amazon is investing heavily in building an intelligent voice assistant in India. Alexa now supports different Indian languages and is integrating with multiple vertical brands in India to accelerate Amazon’s smart audio penetration in the subcontinent. In Spain, digital assistant startup Sherpa.ai is putting its Spanish voice assistants in a range of consumer products through collaborations with the likes of Porsche and Samsung Electronics.

Microsoft meanwhile has announced that its smart voice assistant Cortana will no longer compete with other technology giants, and will instead be repositioned as a complementary service that can be used across different platforms. And so while we won’t be seeing any Cortana smart speakers taking on Alexa or Google Home, Cortana will still show up in a supportive role for Microsoft products such as Office 365 and Azure cloud, following on Microsoft’s integration between Cortana and Alexa last year.

Microsoft’s compromise may be a sensible choice, although the new strategy could also potentially leave Microsoft 365 customers stuck with a Cortana voice interface even though they prefer to interact via Amazon or Google smart speakers.

January 25 – Microsoft Positions Cortana as a Complementary Service

Microsoft recently announces that it will not compete with other players in the voice assistant technology market, but instead position Cortana as a complementary service.
Microsoft is opening Cortana to all competitors and allowing other voice assistants access to its products.
https://www.forbes.com/sites/maciejduraj/2019/01/25/voice-assistant-wars-heat-up-as-microsoft-aims-to-collaborate-with-cortana/#738646b342b7

January 24 – Amazon Taking More Control in IOT

Amazon takes a seat on the Zigbee Board of Directors, will get a say in developing smart home standards
The Zigbee Alliance is an organization of hundreds of companies that create, maintain and deliver open global standards for the Internet of Things and implement the Zigbee wireless protocol.
https://www.theverge.com/2019/1/24/18196463/amazon-zigbee-alliance-board-smart-home-tech-control

Leyi Hu – Analyst, Global Industry

Applying AI to Reduce OverTreatment

Most patients receive an accurate disease diagnosis, but many also can be misdiagnosed, overdiagnosed or overtreated. To address these problems, researchers are exploring ways to apply big data and artificial intelligence to provide patients with effective medical solutions while reducing unnecessary treatment.

Princeton University researchers recently began applying machine learning models to help clinicians more effectively monitor patient condition and identify the best opportunities to conduct lab blood tests for ICU patients. University of Maryland professors meanwhile are experimenting with machine learning algorithms to predict patient response to immunotherapy and reduce the probability of overtreatment situations. A team of researchers at Case Western Reserve University working on the same problem have demonstrated that AI models can predict a patient’s response to chemotherapy.

Current research has also been able to improve determinations on whether breast cancer and lung cancer patients need chemotherapy. The combination of medical data and artificial intelligence is showing great promise towards reducing misdiagnosis and related issues.

January 30 – AI Can Localize Fractures Detected in Wrist X-rays

Researchers from National University of Singapore demonstrate that object detection convolutional neural networks (CNNs) can detect and localize fractures in wrist x-rays with high sensitivity and specificity.
https://pubs.rsna.org/doi/10.1148/ryai.2019180001

January 28 – Machine Learning Helps Identify Cancer Patients for Immunotherapy

Researchers from the University of Maryland, Microsoft Research and Memorial Sloan Kettering Cancer Center are developing a new approach that uses machine learning to identify patients who would benefit from immunotherapy treatment.
https://today.umd.edu/articles/machine-learning-helps-target-cancer-patients-immunotherapy-e3737507-22c7-41b1-a73b-b9da81fad280