Connect with us

Uncategorized

AI could make healthcare fairer—by helping us believe what patients tell us

Published

on

In the last few years, research has shown that deep learning can match expert-level performance in medical imaging tasks like early cancer detection and eye disease diagnosis. But there’s also cause for caution. Other research has shown that deep learning has a tendency to perpetuate discrimination. With a healthcare system already riddled with disparities, sloppy applications of deep learning could make that worse.

Now a new paper published in Nature Medicine is proposing a way to develop medical algorithms that might help reverse, rather than exacerbate, existing inequality. The key, says Ziad Obermeyer, an associate professor at UC Berkeley who oversaw the research, is to stop training algorithms to match human expert performance.

The paper looks at a specific clinical example of the disparities that exist in the treatment of knee osteoarthritis, an ailment which causes chronic pain. Assessing the severity of that pain helps doctors prescribe the right treatment, including physical therapy, medication, or surgery. This is traditionally done by a radiologist reviewing an X-ray of the patient’s knee and scoring their pain on the Kellgren–Lawrence grade (KLG), which calculates pain levels based on the presence of different radiographic features, like the degree of missing cartilage or structural damage.

But data collected by the National Institute of Health found that doctors using this method systematically score Black patients far below the severity of pain that they say they’re experiencing. Patients self-report their pain levels using a survey that asks about pain during various activities, such as fully straightening their knee. But these self-reported pain levels are ignored in preference of the radiologist’s KLG score when prescribing treatment. In other words, Black patients who show the same amount of missing cartilage as white patients self-report higher levels of pain.

This has consistently miffed medical experts. One hypothesis goes that Black patients could be reporting higher levels of pain in order to get doctors to treat them more seriously. But there’s an alternative explanation. The KLG methodology itself could be biased. It was developed several decades ago based on white British populations. Some medical experts argue that the list of radiographic markers that it tells clinicians to look for may not include all the possible physical sources of pain within a more diverse population. Put another way, there may be radiographic indicators of pain that appear more commonly in Black people that simply aren’t part of the KLG rubric.

To test this possibility, the researchers trained a deep-learning model to predict the patient’s self-reported pain level from their knee x-ray. If the resultant model had terrible accuracy, this would suggest that self-reported pain is rather arbitrary. But if the model had really good accuracy, this would provide evidence that self-reported pain is in fact correlated with radiographic markers in the x-ray.

After running several experiments, including to discount any confounding factors, the researchers found that the model was much more accurate than KLG at predicting self-reported pain levels for both white and Black patients, but especially for Black patients. It reduced the racial disparity at each pain level by nearly half.

The goal isn’t necessarily to start using this algorithm in a clinical setting. But by outperforming the KLG methodology, it revealed that the standard way of measuring pain is flawed, at a much greater cost to Black people. This should tip off the medical community to investigate which radiographic markers the algorithm might be seeing, and update their scoring methodology.

“It actually highlights a really exciting part of where these kinds of algorithms can fit into the process of medical discovery,” says Obermeyer. “It tells us if there’s something here that’s worth looking at that we don’t understand. It sets the stage for humans to then step in and, using these algorithms as tools, try to figure out what’s going on.”

“The cool thing about this paper is it is thinking about things from a completely different perspective,” says Irene Chen, a researcher at MIT who studies how to reduce health care inequities in machine learning and was not involved in the paper. Instead of training the algorithm based on well-established expert knowledge, she says, the researchers chose to treat the patient’s self-assessment as truth. Through that it uncovered important gaps in what the medical field usually considers to be the more “objective” pain measure.

“That was exactly the secret,” agrees Obermeyer. If algorithms are only ever trained to match expert performance, he says, they will simply perpetuate existing gaps and inequities. “This study is a glimpse of a more general pipeline that we are increasingly able to use in medicine for generating new knowledge.”

Continue Reading
Comments

Uncategorized

Indonesian logistics startup SiCepat raises $170 million Series B

Published

on

SiCepat, an end-to-end logistics startup in Indonesia, announced today it has raised a $170 million Series B funding round. Founded in 2014 to provide last-mile deliveries for small merchants, the company has since expanded to serve large e-commerce platforms, too. Its services now also cover warehousing and fulfillment, middle-mile logistics and online distribution.

Investors in SiCepat’s Series B include Falcon House Partners; Kejora Capital; DEG (the German Development Finance Institution); Telkom Indonesia’s investment arm MDI Ventures; Indies Capital; Temasek Holdings subsidiary Pavilion Capital; Tri Hill; and Daiwa Securities. The company’s last funding announcement was a $50 million Series A in April 2019.

In a press statement, The Kim Hai, founder and chief executive officer of SiCepat’s parent company Onstar Express, said the funding will be used to “further fortify SiCepat’s position as the leading end-to-end logistics service provider in the Indonesian market and potentially to explore expansion to other markets in Southeast Asia.” SiCepat claims to be profitable already and that it was able to fulfill more than 1.4 million packages per day in 2020.

The logistics industry in Indonesia is highly fragmented, which means higher costs for businesses. At the same time, demand for deliveries is increasing thanks to the growth of e-commerce, especially during the COVID-19 pandemic.

SiCepat is one of several Indonesian startups that have raised funding recently to make the supply chain and logistics infrastructure more efficient. For example, earlier this week, supply chain SaaS provider Advotics announced a $2.75 million round. Other notable startups in the space include Kargo, founded by a former Uber Asia executive, and Waresix.

SiCepat focuses in particular on e-commerce and social commerce, or people who sell goods through their social media networks. In statement, Kejora Capital managing partner Sebastian Togelang, said the Indonesian e-commerce market is expected to grow at five-year compounded annual growth rate of 21%, reaching $82 billion by 2025.

“We believe SiCepat is ideally positioned to serve customers from e-commerce giants to uprising social commerce players which contribute an estimated 25% to the total digital commerce economy,” he added.

Continue Reading

Uncategorized

InsurGrid raises pre-seed financing to help modernize legacy insurance agents

Published

on

Insurance agents spend hours handling paperwork and grabbing client information over the phone. A new seed-stage startup, InsurGrid, has developed a software solution to help ease the process, and make it easier for agents to serve existing clients — and secure new ones.

InsurGrid gives agents a personalized platform to collect information from clients, such as date of birth, driver’s license information and policy declaration. This platform helps agents avoid sitting on long calls or managing back-to-back emails, and instead gives them one spot to understand how all their different clients function. It is starting with property and casualty management.

The startup integrates with 85 insurance carriers, serving as the software layer instead of the provider. Using the InsurGrid platform, insurers can ask clients to upload information and within seconds be registered as a policyholder. This essentially turns into a living Rolodex that insurers can use to access information on the account, and offer quotes on a faster rate.

Image Credits: InsurGrid

There’s a monetary benefit in providing better service. Eden Insurance, a customer of InsurGrid, said that people who submit information through the platform converted at an 82% higher rate than those who don’t. Jeremy Eden, the agency owner of Eden Insurance, said they were able to show consumers that its plan was $300 cheaper than its existing rate.

At the heart of InsurGrid is a bet from the founding team that legacy insurance agents aren’t going anywhere. Co-founder/CEO Chase Beach pointed out that the majority of the $684 billion of annual property and casualty insurance premiums in the United States is distributed by approximately 800,000 agents working in 16,000 brokerages. So far, InsurGrid works with more than 150 of those agencies.

When asked if InsurGrid ever had plans to offer its own insurance, similar to insurtech giants Hippo, Lemonade and Root, Beach said that it is solely working on innovating around the sales process for now. He said that these big companies, which have either recently gone public or are planning to, still rely on agents to be successful.

“Instead of us replacing the insurance agent, what if we gave them that same level of technology of a Hippo or large carrier,” Beach said. “And provide them with the digital experiences so they can compete in 2021.”

As time goes on, he sees insurance agents taking the same role that financial advisors or real estate agents take: “very much involved in the process because they are that expert.”

Other startups that have popped up in this space include Gabi, Trellis and Canopy Connect. The differentiator, the team sees, is that Beach comes from a 144-year-old insurance legacy, giving him key insights on how to sell to agents in a successful and effective way. It is starting with sales, but expect InsurGrid to expand to other parts of the insurance process as well.

To help them compete with new and old startups, InsurGrid recently raised $1.3 million in pre-seed financing to help it fulfill its goal to be the “underdog for the underdogs,” Beach said. Investors include Engineering Capital, Hustle Fund, Vess Capital, Sahil Lavingia and Trevor Kienzle.

Continue Reading

Uncategorized

Backed by Blossom, Creandum and Index, grocery delivery and dark store startup Dija launches in London

Published

on

Dija, the London-based grocery delivery startup, is officially launching today and confirming that it raised £20 million in seed funding in December — a round that we first reported was partially closed the previous month.

Backing the company is Blossom Capital, Creandum and Index Ventures, with Dija seemingly able to raise pre-launch. In fact, there are already rumours swirling around London’s venture capital community that the upstart may be out raising again already — a figure up to £100 million was mooted by one source — as the race to become the early European leader in the burgeoning “dark” grocery store space heats up.

Image Credits: Dija

Over the last few months, a host of European startups have launched with the promise of delivering grocery and other convenience store items within 10-15 minutes of ordering. They do this by building out their own hyper-local, delivery-only fulfilment centres — so-called “dark stores” — and recruiting their own delivery personnel. This full-stack or vertical approach and the visibility it provides is then supposed to produce enough supply chain and logistics efficiency to make the unit economics work, although that part is far from proven.

Earlier this week, Berlin-based Flink announced that it had raised $52 million in seed financing in a mixture of equity and debt. The company didn’t break out the equity-debt split, though one source told me the equity component was roughly half and half.

Others in the space include Berlin’s Gorillas, London’s Jiffy and Weezy, and France’s Cajoo, all of which also claim to focus on fresh food and groceries. There’s also the likes of Zapp, which is still in stealth and more focused on a potentially higher-margin convenience store offering similar to U.S. unicorn goPuff. Related: goPuff itself is also looking to expand into Europe and is currently in talks to acquire or invest in the U.K.’s Fancy, which some have dubbed a mini goPuff.

However, let’s get back to Dija. Founded by Alberto Menolascina and Yusuf Saban, who both spent a number of years at Deliveroo in senior positions, the company has opened up shop in central London and promises to let you order groceries and other convenience products within 10 minutes. It has hubs in South Kensington, Fulham and Hackney, and says it plans to open 20 further hubs, covering central London and Zone 2, by the summer. Each hub carries around 2,000 products, claiming to be sold at “recommended retail prices”. A flat delivery fee of £1.99 is charged per order.

“The only competitors that we are focused on are the large supermarket chains who dominate a global $12 trillion industry,” Dija’s Menolascina tells me when I ask about competitors. “What really sets us apart from them, besides our speed and technology, is our team, who all have a background in growing and disrupting this industry, including myself and Yusuf, who built and scaled Deliveroo from the ground up”.

Menolascina was previously director of Corporate Strategy and Development at the takeout delivery behemoth and held several positions before that. He also co-founded Everli (formerly Supermercato24), the Instacart-styled grocery delivery company in Italy, and also worked at Just Eat. Saban is the former chief of staff to CEO at Deliveroo and also worked at investment bank Morgan Stanley.

During Dija’s soft-launch, Menolascina says that typical customers have been doing their weekly food shop using the app, and also fulfilling other needs, such as last minute emergencies or late night cravings. “The pain points Dija is helping to solve are universal and we built Dija to be accessible to everyone,” he says. “It’s why we offer products at retail prices, available in 10 minutes – combining value and convenience. Already, Dija is becoming a key service for parents who are pressed for time working from home and homeschooling, as one example”.

Despite the millions of dollars being pumped into the space, a number of VCs I’ve spoken to privately are sceptical that fresh groceries with near instant delivery can be made to work. The thinking is that fresh food perishes, margins are lower, and basket sizes won’t be large enough to cover the costs of delivery.

“This might be the case for other companies, but almost everyone at Dija comes from this industry and knows exactly what they are doing, from buying and merchandising to data and marketing,” Menolascina says, pushing back. “It’s also worth pointing out that we are a full-stack model, so we’re not sharing our margin with other parties. In terms of the average basket size, it varies depending on the customer’s need. On one hand, we have customers who do their entire grocery shop through Dija, while on the other hand, our customers depend on us for emergency purchases e.g. nappies, batteries etc.”

On pricing, he says that, like any retail business, Dija buys products at wholesale prices and sells them at recommended retail prices. “Going forward, we have a clear roadmap on how we generate additional revenue, including strategic partnerships, supply chain optimisation and technology enhancements,” adds Menolascina.

Dija testing on Deliveroo

Image Credits: TechCrunch

Meanwhile, TechCrunch has learned that prior to launching its own app, Dija ran a number of experiments on takeout marketplace Deliveroo, including selling various convenience store items, such as potato chips and over-the-counter pharmaceuticals. If you’ve ever ordered toiletry products from “Baby & Me Pharmacy” or purchased chocolate sweets from “Valentine’s Vows,” you have likely and unknowingly shopped at Dija. Those brands, and a number of others, all delivered from the same address in South Kensington.

“Going direct to consumer without properly testing pick & pack is a big risk,” Menolascina told me in a WhatsApp message a few weeks ago, confirming the Deliveroo tests. “We created disposable virtual brands purely to learn what to sell and how to replenish, pick & pack, and deliver”.

Continue Reading

Trending