DIGITAL HEALTH BRIEFING: Alphabet, Apple, Microsoft patents reveal health interest - NHS staff in trouble for using chat apps - Google uses expert opinion to improve AI
Welcome to Digital Health Briefing, the newsletter providing the latest news, data, and insight on how digital technology is disrupting the healthcare ecosystem, produced by Business Insider Intelligence.
Have feedback? We'd like to hear from you. Write me at: firstname.lastname@example.org
TECH GIANTS' PATENT FILING HIGHLIGHT HEALTHCARE INTEREST: Alphabet, Apple, and Microsoft filed 313 healthcare patents between 2013 and 2017, according to Ernst and Young. As the healthcare industry looks to digital initiatives to improve and enhance care delivery and hospital operations, these tech companies are putting their engineering skills, and data and analytics expertise to use to help solve some of healthcare's biggest issues, such as interoperability troubles and data analysis.
Here's an overview of where the interests of Alphabet, Apple, and Microsoft lie in healthcare, based on patent activity:
- Alphabet's patent activity was primarily concentrated on its AI company DeepMind and Verily Life Sciences, its health-focused business. The company's health initiatives include exploration into chronic illnesses like diabetes, the use of bioelectronics, and "smart" operating rooms.
- Apple is working to turn its consumer products into healthcare devices. For instance, a recent update to iOS 11 transformed the Health app into an electronic medical record (EHR) store. And the company partnered with Stanford University to explore how the Apple Watch could be used to identify atrial fibrillation (aFib).
- Microsoft's focus is on expanding its AI capabilities and developing monitoring devices for chronic conditions. Microsoft's Healthcare NeXT initiative, for example, was launched in 2017 with the aim of accelerating innovation in the healthcare industry by using advancements in AI and cloud computing, such as its partnership with Adaptive Biotechnologies to build a universal blood test that's capable of screening for dozens of diseases at once.
The tech giants' platform model and broad reach gives them an advantage over smaller niche players in the market. Clients can use the tech companies' platforms to offer a range of services for a distinct condition, such as diabetes, at scale, taking advantage of their consumer reach, while also leveraging cloud computing for things like data analysis, forms of AI, and server storage.
Enjoy reading this briefing? Sign up and receive Digital Health Briefing to your inbox.
NHS STAFF PUNISHED FOR USING MESSAGING APPS TO SPEAK TO PATIENTS: Around 500,000 staff in the UK National Health Service (NHS) are using instant messaging (IM) apps, such as WhatsApp and Facebook Messenger, to communicate, due in part to frustration with NHS-approved communication tools, according to a report by mobile tech company CommonTime. Further, 38% of NHS staff use an IM at least once each day for things like roster management, referrals, second opinions, and shift handovers. The staff found that IMs made communication faster, improved patient engagement, and were easier to use than those offered by their organizations. The issue is that these messaging apps aren't compliant with data protection policies and in some instances led to confidentiality threats and alleged malicious use, such as sharing patient data on social media. Now, 29,000 NHS employees are facing disciplinary action for using non-approved IMs. The need for secure communication tools is particularly important in the healthcare industry, which regularly suffers data breaches, largely caused by a lack of compliance to best practices, according to a new McAfee report. Organization-approved communication can help mitigate the risk of staff accidentally sending patient information to external sources. However, to get staff using the tools, the NHS will need to improve the user experience.
GOOGLE RESEARCHERS IMPROVE AI ACCURACY USING PANEL OF EXPERTS: Google researchers have greatly improved the accuracy of their diabetic retinotherapy (DR) algorithm by training it with a small set of images adjudicated by eye experts, MobiHealthNews reports. Google researchers developed the algorithm, which was unveiled in December 2017, to help identify cases of DR - a vision-threatening disease caused by high blood-sugar levels damaging blood vessels in the retina - and similar eye diseases, such as glaucoma and age-related macular degeneration. In its first round of testing, the machine learning technology was able to identify cases of DR with a sensitivity of 90.5% and specificity of 91.6% - sensitivity refers to the ability to correctly identify people with the disease, whereas specificity is the ability to correctly identify people without the disease. However, using the new adjudicated images, which were graded by a consensus of US board-certified ophthalmologists and retinal specialists before being fed to the AI, researchers were able to increase the algorithm's sensitivity to 97% and its specificity increased to 92%. Moreover, the AI outperformed the majority decision of ophthalmologists, who had an overall sensitivity of 84% and specificity of 98%.
The vast improvements to the algorithm's accuracy, particularly in comparison to the experts' results, "raises the bar for reference standards the field of applying machine learning to medicine," according to Google AI Research Group product manager Lily Peng, MD, PhD. Google's not alone in its endeavor to automate the process of identifying DR. Researchers at Madras Diabetes Research Foundation in India found that a smartphone app, called Remido 'Fundus on phone', coupled with AI, had 95% sensitivity and 80% specificity for detecting DR. And Iowa-based company IDx announced in February that its AI-powered tool IDx-DR is undergoing expedited review by the US Food and Drug Administration.
AI-POWERED SOFTWARE COMPANY TO MAKE DERMATOLOGY SOLUTION: Proscia, a company that develops AI-powered software for clinics, is partnering with one of the largest dermatopathology labs in the US to help the company develop its AI image analysis pathology solution for a Q4 2018 launch. Proscia will use the significant data from the lab, including pathology slides and expert pathologists' ground truths - information provided by direct observation -to "teach" its AI and improve its ability to recognize diseases with more accuracy. While not a diagnostic tool, pathologists will be able to use Proscia's platform for clinical decision support (CDS), by increasing diagnostic confidence for faster lab results. The solution will help to speed up the diagnostic process for around 70% of dermatopathology case volume. As the medical specialist and care provider workforce declines, solutions that can automate at scale will become increasingly important. Harnessing machine learning capabilities for things like image analysis within the pathology field will help streamline the pathology and diagnosis process, enabling specialists to improve efficiency and enhance productivity. It could also help cut back on mistakes caused by fatigue from labor-intensive assessments.