Ada is an AI-powered doctor app and telemedicine service

Ada, a London and Berlin-based health tech startup, sees its official U.K. push today, and in doing so joins a number of other European startups attempting to market something akin to an AI-powered ‘doctor’.

The company’s mobile offering bills itself as a “personal health companion and telemedicine app” and via a conversational interface is designed to help you work out what symptoms you have and offer you information on what might be the cause. If needed, it then offers you a follow up remote consultation with a real doctor over text.

In a call, two of Ada’s founders — CEO Daniel Nathrath and Chief Medical Officer Dr Claire Novorol — explained that the app has been six years in the making, and actually started life out as being doctor-facing, helping clinicians to make better decisions. The same database and smart backend is now being offered to consumers to access, albeit with a much more consumer-friendly front-end.

In my brief testing of the app, I plugged in the symptoms of a sore or red eye. After drilling through a quite extensive set of questions, many of which appeared to relate to the answers I’d previously given, the Ada app provided three possible conditions, and advised that they could be successfully treated at home.

That, say the company’s founders, reflects one of the main benefits of an AI-driven healthcare app like Ada, which is to empower patients to make more informed decisions about their health. Or, to out it more bluntly, to ensure we only visit a doctor when we need to and, more generally, can be proactive in our healthcare without adding the need for greater human doctor resources.

In other words, just like competitor Babylon, which has added its own AI-powered triage functionality and is backed by two of DeepMind’s founders, this is about using technology to help healthcare scale.

“Ada has been trained over several years using real world cases, and the platform is powered by a sophisticated artificial intelligence (AI) engine combined with an extensive medical knowledge base covering many thousands of conditions, symptoms and findings,” explains the company.

“In every assessment, Ada takes all of a patient’s information into consideration, including past medical history, symptoms, risk factors and more. Through machine learning and multiple closed feedback loops, Ada continues to grow more intelligent, putting Ada ahead of anyone else in the market”.

With that said, Ada isn’t claiming to replace your doctor anytime soon. Like a lot of AI being applied to various verticals, not least healthcare, the app is designed to augment the role of humans, not replace it altogether.

This happens very tangibly in two ways: helping to act as a prescreen consultation before, if needed, being handed off to a real doctor for further advice, or simply helping to create a digital paper trail before a consultation takes place. By getting some of the most obvious symptom-related questions out of the way and captured and analysed by the app, it saves significant time during any follow up consultation.

Novorol tells me that since the app went live, feedback has already shown it to successfully diagnose both common and quite rare conditions. She also talked up the notion that Ada’s AI, since it has and continues to be trained by real doctors, essentially pools a lot of shared expertise. It did start off as a tool to help doctors avoid misdiagnosis, after all.

I asked how Ada compares to Babylon, and although he slightly comically refused to say the company’s name out loud, CEO Nathrath said that unlike competitors, AI isn’t an afterthought. Where others have started with a ‘Skype your doctor’ type offering and added AI, Ada is six years AI in the making and is only now adding remote consultations.

Meanwhile, the startup is being quite secretive regards how it is funded. Aside from an EU grant, Ada Health is said to be backed by unnamed private individuals.

Xiaomi’s Mi 6 puts iPhone 7 camera tech into a $360 phone — and there’s no headphone jack

Xiaomi unveiled its newest flagship smartphone — the Mi 6 — today at an event in Beijing, having skipped Mobile World Congress, the telecom industry’s largest annual event where it announced the Mi5 last year.

There’s plenty at stake with this new device, given that Xiaomi suffered a sales slump last year with CEO Lei Jun admitting that the company has entered a transitional period after growing too fast.

So, what about the Mi 6?

Well, the first thing to note is that there are plenty of similarities to the iPhone 7, but price isn’t one of them. The entry model — featuring 64 GB of storage — comes in at 2499 RMB, that’s around $360, with a 128 GB option (2899 RMB, $420) and ceramic edition (2999 RMB, $435) completing the range. All three are far cheaper than iPhone equivalents, but, interestingly for Xiaomi, the range is more expensive that the company’s usual flagship prices.

The most obvious iPhone comparison is that there is no headphone jack on the Mi 6, just as Apple elected with last year’s iPhone 7. Whether the Chinese company is following the trend, or doing what makes sense for itself in this instance, that is already ammo for Xiaomi skeptics.

But that’s not all. In a further parallel, the 5.5-inch Mi 6 includes a 12-megapixel dual rear camera — à la the iPhone 7+ — which mixes a wide-angle lens and a telephoto lens. Like the iPhone, the Mi 6 includes a bokeh-style photography option, alongside 10x digital zoom, 2x lossless zoom, and optical image stabilization (OIS) technology. Also on board is an under-glass fingerprint sensor on the front. Xiaomi shipped a similar sensor in the Mi5s Plus last September and, if you believe leaks reported this week, Apple may have similar plans for its next iPhone.

A sample photo of the Mi 6 ‘portrait’ mode supplied by Xiaomi

With the Mi 6, Xiaomi has bumped up its RAM to 6GB, the most it has ever offered in a smartphone. The device is powered by a Snapdragon 835 10nm processor with a 64-bit, octa-core CPU with a whopping 3350 mAh battery that the company said will last a day thanks to “optimization” controls built into its MIUI operating system. Xiaomi announced its own-designed processor in February but it isn’t present here. The chip — called the Surge S1 — made its debut in the Mi5c two months ago but it will take some fine-tuning before it is part of a flagship.

Visually, the device is a departure from the Mi5 with curved edges that Xiaomi said are made possible through its use of “four-sided 3D glass,” while there is a new silver color option that includes a mirrored finish and that rather fetching ceramic model. Other notable features include dual speakers for stereo audio, improved 2.2 dual Wi-Fi technology, and new screen options that include a new night display and reduce blue ray output.

The phone goes on sale from Mi.com and Mi Home Stores in China on April 28. An international expansion to “selected” markets will come later, Xiaomi said, although it didn’t provide specific dates for that.

 

 

Is this new package enough to regain ground that Xiaomi lost in China?

Xiaomi appears to be seeking a hailo effect with a new flagship that is packed full of tech which, on paper at least, compares favorable to the iPhone and Samsung’s Galaxy S8. The company has always sold higher volumes with its lower priced phones — its Redmi range is priced around $150 — so expect the Mi 6 to be a mark in the sand for what its technology and manufacturing chops are capable of. (Although it opted not to innovate on the near-bezel-less Mi Max that won critical acclaim last year.)

The challenge facing the company is tough. In 2016, rivals Huawei, Vivo and Oppo were among the challengers that rose up with competitively priced phones and strong offline distribution reach to knock Xiaomi off the number one perch for smartphones sales in China. Lei Jun has promised to bounce back and expand Xiaomi’s offline commerce efforts, while he intends to continue its ‘smart device’ ecosystem push to increase customer engagement and generate increased revenue. Indeed, the Xiaomi CEO recently likened his firm to U.S. retailer Costco rather than Apple, the latter being a common comparison made in previous years.

View image on Twitter

 Follow

Bryan Ma @bryanbma

If you're covering #XiaomiMi6, Xiaomi was at its peak in China smartphones in 2Q15 (#1 with 16% share). By 4Q16, Xiaomi was #5, 7.4% share

11:33 AM - 19 Apr 2017 · Central Region, Singapore

 
•  44 Retweets
 
•  44 likes

Overseas, it remains unclear how Xiaomi’s internationalization effort is going, particularly following the departure of Hugo Barra, who led the project but has since move on to Facebook. Xiaomi revealed it cleared $1 billion in revenue in India, its second largest market behind China, last year while Lei Jun added today that it ranks second in the country, but nothing has been said of its performance in the other 20-odd countries where its phones are sold.

Facebook open sources Caffe2, its flexible deep learning framework of choice

Today Facebook open sourced Caffe2. The deep learning framework follows in the steps of the original Caffe, a project started at the University of California, Berkeley. Caffe2 offers developers greater flexibility for building high-performance products that deploy efficiently.

This isn’t the first time that Facebook has engaged with the Caffe community. Back in October, Facebook announced Caffe2Go, what effectively was a mobile CPU and GPU optimized version of Caffe2 (they even both have Caffe2 in their names if you parse it right). Caffe2Go received attention at that time because its release coincided with Style Transfer.

Notably, the company also released extensions to the original Caffe. The majority of these changes make Caffe more attractive to developers building services for large audiences. For projects where resources are of no consequence, Facebook has historically turned to Torch — a library it finds optimal for research use cases.

Every tech company wants to tout the scalability of its machine learning framework of choice. I asked Yangqing Jia, the lead author on Caffe2, what he thought of MXNet and the noise Amazon has been making about its ability to scale. Reasonably, he was cautious about dropping benchmarking numbers for comparison. These numbers can have meaning, but they are heavily influenced by the actual implementation of a machine learning model and subject to a fair amount of “DIY” volatility.

Yangqing Jia, the lead author on Caffe2 and Alex Yu, leader of business development

“All frameworks are more or less at a similar scalability factor,” explained Jia. “We’re pretty confident that Caffe2 is probably a little bit better than the rest.”

Facebook is pouring a lot of resources into both Caffe2 and PyTorch. Today’s release accompanies partnerships at the hardware, device and cloud levels. Alex Yu, leader of business development for Caffe2, explained to me that Facebook aimed to include the market leaders in each category. This meant Nvidia and Intel on the hardware side, Qualcomm on the device side and Amazon and Microsoft on the cloud side. And while Google wasn’t targeted, a GCP partnership wouldn’t be out of the question going forward.

Prior to release, Caffe2 was deployed at scale across Facebook. The team also took considerations for the developer communities familiar with the original Caffe. Caffe models can be easily converted to Caffe2 models with a utility script. Facebook is releasing documentation and tutorials and has put Caffe2’s source code on GitHub.

Instagram on Android gets offline mode

80% of Instagram’s users 600 million users are outside the US, so it needed a way to provide a better experience for users with limited network connectivity or no data plan.

Today at F8, Instagram announced it’s built support for using most of its features without Internet access. Much of this functionality is now available on Android, which is the preferred device type in the developing world. More will come in the following months, and Instagram tells me its exploring an iOS version.

Instagram engineer Hendri says offline users will be able to see content previously loaded in Instagram’s feed. People can leave comments, Like things, save media, or unfollow people — all of which will go through when they reconnect. Profiles they’ve visited before will be visible, as will old versions of the Explore tab or their own profile.

The engineering gymnastics required to do this could help Instagram grow in developing nations where data is either too expensive for everyone to afford, or there aren’t omnipresent or stable data connections. Facebook’s developing world app Facebook Lite shot to 200 million users in just a year, proving the big opportunity Instagram could seize by allowing users to enjoys the app even in isolation. While Snapchat seems to have forgotten about the developing world, Instagram knows everyone everywhere wants visual communication.

CLEAR raises $15 million to expand biometric security

CLEAR, the biometric security firm popular in U.S. airports and stadiums, has raised $15 million from T. Rowe Price to expand their footprint. The company, led by CEO Caryn Seidman-Becker, is already partially owned by Delta Airlines.

“We are obsessed with our customers, and can’t wait to bring them more ways to use CLEAR. We lead the industry in secure biometric-powered customer experiences and the opportunities to expand into new verticals are endless,” said Seidman-Becker. “CLEAR is just scratching the surface of its growth potential, and we’re proud to have wonderful, long-term partners in T. Rowe Price to help us get there.”

CLEAR’s numbers are rising. There are 1 million members using biometrics at 21 airports, as well as six baseball stadiums and one NBA arena. LAX in Los Angeles will be the next airport to activate the technology and member enrollment is up 150 percent since last year.

The financing will move CLEAR into “new verticals” and add more airports and stadiums to the network. The company hopes to bring its biometric technologies to create seamless transfers and payments.

“We see a frictionless future where people no longer need to carry cash, credit cards or IDs with them,” said Seidman-Becker. “Our simple tap and go experience will fundamentally change the way we live, work and interact with the world.”

Streamers flock to YouTube Live, but the money (and crowd) is still at Twitch

YouTube Live is making impressive strides in catching up to incumbent Twitch, but it has a long way to go yet. On the bright side, it doesn’t look like they’ll run out of road: Streaming looks to be a fairly sustainable business, suggest stats from Streamlabs’ latest report.

According to data gleaned through Streamlabs’ streaming assistant app, YouTube Live has outpaced Twitch in growth over the last six months. That’s to be expected with a newish service offered by a powerful competitor, but it’s also not uncommon to see challengers like this fail to gain serious traction at all.

Twitch these days has somewhere north of 250,000 monthly active streamers, and that’s growing in fits and starts; 5,000 this month, 15,000 the next. YouTube added 23,000 its first month, and has grown to around 75,000 (again, monthly actives). That’s healthy growth and its momentum seems far from spent.

That said, Twitch still generated 96 percent of the tip revenue that Streamlabs (which is used by huge numbers of streamers) directly observed.

One pleasantly surprising statistic newly reported is that longtime stream viewers tip considerably more than new ones. Two-year-old accounts tipped more than $80 total on average, while brand new accounts tipped around $23. The numbers smoothly increase between those extremes, implying this is something people are adopting as a regular hobby, just like anything else, and spending money accordingly.

That’s supporting the streamers themselves, who are increasingly likely to pull real income from their channel. The number of non-sponsored streamers is increasing, as well.

I chatted briefly with Ali Moiz, CEO of Streamlabs, about the state of the industry. In particular, I was afraid that with Twitch no longer growing like a weed (merely like a regular plant), that could put some dampers on the wild speculation about the future of the streaming business.

Moiz acknowledged that Twitch’s growth had slowed, but pointed out that was more due to healthy competition.

“For a new streamer starting today there’s a much greater choice than there was a few years ago. Back then basically anyone interested in live streaming had to go to Twitch,” he said. Now you have YouTube, Facebook, Instagram and half a dozen others.”

But, I asked, more options doesn’t mean more streamers or more viewers necessarily. Where’s the next five years of growth going to come from?

“In my view the biggest chunk of growth is going to come from outside of gaming. Mobile broadcasting is just getting started,” Moiz said. “Travel, fashion, beauty, concerts, food (that’s big in Korea). Gaming is going to have nice incremental growth, but it’s not going to double or triple like it has in the past.”

You can read the full report (they come out quarterly now, since the first one in January) over at Streamlabs’ Medium post.

Troubled travel startup Zozi gets acquired by Peek

Zozi, the drama-filled travel industry startup whose former CEO recently sued its board, has been acquired by Peek, a rival in the tours and activities business where Zozi also competes. Terms of the deal were not disclosed, but the two companies will be joining forces, with the assets and team from Zozi heading over to Peek.

The news was reported by Skift, which notes that Peek is basically hiring the Zozi employees and bringing its customers on board, while the Zozi brand and marketplace will be phased out.

This is a nice, soft landing for troubled Zozi, which has been struggling with legal battles following the firing of founder and CEO T.J. Sassani, who sued the board for wrongful termination and a host of other outrageous claims. Meanwhile, Zozi staff told us that Sassani had been largely absent from the company for months (even skipping its holiday party), had difficult relations with vendors and was spending money on ventures that other members of the executive team didn’t believe in and thought could potentially bankrupt the business.

In recent days, Zozi had shifted its focus away from its travel and tours marketplace to Zozi Advance, its B2B platform for bookings and reservations. The changes forced the company to lay off nearly 40 percent of its staff earlier this year.

Peek, on the other hand, grew 10x in the last two years, it says. Last summer the company announced a $10 million round of funding, which included Trulia founder Pete Flint joining its board. It has used that funding to invest in its bookings platform, Peek Pro.But Zozi Advance held promise, the company believed. It said that the SaaS product allowed it to grow its revenue nearly 3x in 2016, and was on track to double its revenue this year.

Zozi tells us that its Zozi Advance customers will be brought onto Peek Pro’s platform in the months ahead. The merchants on the Zozi marketplace will be moved over to Peek.com, but Zozi’s brand and consumer offering will later be shuttered.

Not all Zozi employees will be joining Peek, from the sound of things. The company says it’s currently finalizing the team structure, with plans to bring together “key members” of the Peek and Zozi teams to “leverage the best talent in the space.”

“Peek shares our vision around putting operators first, said Elon Boms, Zozi board member and investor, about the deal. “And above all, we care about making sure we provide our operators with the best service and technology to help grow and manage their business. This is a big win for the industry and a big win for our operators.”

HP’s new x360 laptops miss the point of 4K on the Spectre but get security right with the EliteBook

The Spectre 13 x360 that I strangely loved from late last year is back with a 4K screen and pen support for the Windows Creators update that went global this month. There’s also a new EliteBook x360 for enterprise users, with beefed-up security via a fingerprint scanner, Windows Hello and always-on security monitoring for the BIOS.

One x360 laptop is better for work, the other for multimedia. Both are some of the best Windows machines out there, but with a few caveats.

A secure laptop is only as secure as the network it’s on

HP EliteBooks are often sold in contracts to corporations or startups that need Windows systems in bulk for their employees. The EliteBook x360 is no different, except we haven’t seen the x360 hinge with a pen before on an EliteBook. The result is a work machine that can be laid flat with its dual microphones for a conference call, but can transform into a tablet with pen support — reminiscent of the Microsoft Surface experience. This functionality is assisted by the Windows Creators update that added Paint 3D, Windows Ink in Photos/Edge and a whole suite of features I previously covered.

Specs-wise, the EliteBook x360 is similar to its consumer-oriented counterpart. A Core i7 processor, 16GB of RAM, 512GB SSD, a sharp 1080p HD screen and lots of legacy ports — HDMI-out, microSD, secure card, Kensington lock slot, etc. — are all included.

The difference between the EliteBook and the Spectre is most notably: security.

But if it’s just a powerful Windows laptop you’re looking for, then the Spectre will suit you just fine.

The difference between the EliteBook and the Spectre is most notably: security. Every 15 seconds the BIOS is scanned for tampering. If detected, the EliteBook shuts down and installs a fresh copy of the BIOS. Of course, the whole system is also encrypted and supported by Windows 10 Pro.

1. 

 

 

There are also quite a few ways to lock the system: Windows Hello facial recognition (via infrared camera), a fingerprint sensor or the HP WorkWise app (iOS/Android) that acts as a proximity lock, like the key fob to a Tesla or similar luxury car.

Battery life is solid, ranging from 12-15 hours in my personal use.

However, your system is really only as secure as your connection to the internet and the services you use. A VPN, good judgement of networks, sites, files and personal network security (or the corporate network managed by an IT manager) is what will decide whether your system gets compromised or not.

But fret not: If all of that physical access security isn’t enough, HP partnered with 3M to start shipping EliteBooks with an optional SureView privacy filter. It activates 35 degrees from the center viewpoint, getting brighter, not darker to provide a privacy filter. The downside is you’ll take an hour battery life penalty, but the trade-off is it can be easily activated from the keyboard.

4K and pen support — that’s it?

I’m torn with the new Spectre x360: I recommended it wholeheartedly last year, as did a number of other publications. It’s been less than six months and now there’s a 4K option with the same internals.

In the real world, having a 4K screen doesn’t offer that much of an advantage over the original model — other than limited high-res content consumption, or if you just like pixels. To top it off, the 13-inch 4K screen doesn’t adhere to any color-grading standard (like Adobe’s, for example), which would make it great for multimedia editing.

So it really is just a pretty face, driven by otherwise compelling hardware.

The regular full-HD panel looks like an Etch-A-Sketch in comparison.”

However, the quality of the visual upgrade is highly noticeable. The regular full-HD panel looks like an Etch-A-Sketch in comparison. When buying a new laptop like this, you’ll need to ask yourself if trading a few hours of battery life for 3840 x 2160 pixels is worth it — in this case, it really isn’t.

While I used to proudly reach 10 hours of usage on the HD Spectre x360, I can only get by 6-7 hours on the 4K model. Time I could have spent working or watching (with content) on a less crisp, but still vibrant and sharp HD screen.

In summary, I’d pass on the 4K for the consumer x360 and go for extra security on the enterprise x360.

Prices as reviewed: $1,269+ for EliteBook x360, $1,599 for Spectre x360 with 4K.

Audi’s latest concept is a new all-electric Tesla Model X competitor

Audi’s newest electric vehicle is a crossover SUV with a 4.5-second 0-60mph time, and an impressive range of around 300 miles when its 95kWh battery is topped up. The Audi e-tron Sportback is set for production starting in 2019, and it’s part of Audi’s larger strategy to bring at least three fully electric cars to market by 2020, and to have a quarter of its overall lineup equipped with EV batteries by 2025.

The carmaker is pursuing electric options fairly aggressively, likely part of an overall shift dictated by its parent company Volkswagen, which has been more emphatic about its EV plans since its diesel emissions scandal. The e-tron Sportback definitely sounds like it could be a strong contender to help show that the electric strategy is full of potential for a range of buyers, and it would be nice to see a vehicle that can compete in many ways with Tesla’s gull-wing luxury electric SUV, the Model X.

Audi has already announced the all-electric e-tron quattro, which is set to begin being offered for sale in 2018, and by then, not to mention when the e-tron Sportback hits the market, there should be no shortage of options in the high-end electric SUV market, judging by the announced roadmaps of a number of carmakers.

Futuristic tech touches make the Audi a sight to behold, including Matrix LED, laser lighting and OLEDs designed to provide new levels of light-based communication with the outside world, as well as passengers within. Based on this and other concepts we’ve seen, if you’re not crazy about the idea of cars becoming “expressive” anthropomorphized futurist love bugs, the future isn’t for you.

The e-tron Sportback also all but does away with blind spots thanks to strategically positioned cameras that pipe real-time images to displays built into the interior of the car’s doors, giving you a view as to what’s going on all around.

Facebook announces React Fiber, a rewrite of its React framework

Facebook has completely rewritten React, its popular JavaScript library for building user interfaces. The company hasn’t previously talked much about React Fiber, as the project is called, but it has actually been working on it for a while. It’s now ready to talk about this project publicly in more detail (after word about it started spreading last year) and the plan is to put this rewrite into the hands of developers once React 16.0 launches later this year. It’s already in use on Facebook.com today, which clearly indicates that Facebook itself thinks it’s ready for prime time.

In addition, it is also launching a rewrite of Relay, its framework for building data-heavy applications.

React Fiber

The idea behind React Fiber, the company tells me, is to take what the company has learned from developing React the first time around and put that into an updated framework that is still fully backwards compatible with existing React-based applications. React Fiber, Facebook tells me, will become the foundation of any future improvements and feature development of the React framework.

The main focus here was to make React as responsive as possible, Facebook engineer — and member of the React core team — Ben Alpert told me in an interview earlier this week. “When we develop React, we’re always looking to see how we can help developers build high-quality apps quicker,” he noted. “We want to make it easier to make apps that perform very well and make them responsive.”

In light of this theme, it’s no surprise that the highlights of this new release are built-in primitives for scheduling and incremental rendering. “We want to make sure we render the right stuff at the right time,” Alpert said, and added that “responsiveness was a huge push here.”

But why rewrite React from scratch? “It was not necessarily that the old code base was bad, but we wanted to start with a new foundation that could power everything we do going forward,” Alpert said. That means the new code was developed from the ground up to be extensible, for example.

Alpert stressed that React Fiber will be backward compatible, though as with all major React updates, there will be a few small breaking changes. The team says it doesn’t anticipate that these will be problematic for developers, though. “We always had a strong API contract, so that gives us the flexibility to reimplement,” he added.

Relay Modern

As Facebook also today announced, Relay — the company’s JavaScript framework for building data-driven applications — has also been rewritten with a similar emphasis on performance and extensibility. Relay combines React with Facebook’s GraphQL query language and now Relay Modern, as the company calls this rewrite, is meant to push this concept further and overcome some of the limitations of the original design. That also meant simplifying some of the design to enhance the overall performance of the framework. “Relay Modern retains the best parts of Relay — colocated data and view definitions, declarative data fetching — while also simplifying the API, adding features, improving performance, and reducing the size of the framework,” the team explains in today’s announcement. To do this, the team implemented a number of changes, but most importantly, it adopted static queries and ahead-of-time optimizations.

Static queries essentially ensure that complex queries that aren’t altered by runtime conditions can be pre-built and offloaded to Facebook’s servers. So instead of sending complex queries across the network, all an application has to send is a string that identifies the pre-set query and the variables needed to complete it. Related to this, the ahead-of-time optimization feature in the Relay compiler now looks at the query structure to optimize the query that is now stored on the server to execute it faster — and hence return results to the user faster, too. Other new features in React Modern include built-in garbage collection, for example.

For developers who are already using an older version of Relay, Relay Modern comes with a compatibility API.

Facebook says that when its teams switched the Marketplace tab in the Facebook app from Relay to Relay Modern, the time to interaction on Android improved by an average of 900ms. While that doesn’t sound like much, every second on mobile counts, and that’s enough to make an application feel noticeably more responsive than before.