Connect with us


The convergence of deep neural networks and immunotherapy



The convergence of deep neural networks and immunotherapy

What do deep neural networks and cancer immunotherapy have in common?

While both are among the most transformational areas of modern science, 30 years ago, these fields were all but ridiculed by the scientific community. As a result, progress in each happened at the sidelines of academia for decades.

Between the 1970s and 1990s, some of the most prominent computer scientists, including Marvin Minsky, in his book “Perceptrons,” argued that neural networks (the backbone of most modern AI) would never work for most applications. He exposed flaws in the early conceptions of neural networks and argued that the whole approach was ineffective.

Meanwhile, during the 1980s through the 2000s, neural network pioneers and believers – Geoffrey Hinton, Joshua Bengio, and Yann LeCun – continued their efforts and pursued their intuition that neural networks would succeed. These researchers found that most of the original ideas were correct but simply needed more data (think of ImageNET), computational power and further modeling tweaks to be effective.

Geoffrey Hinton, Yoshua Bengio, and Yann LeCun were awarded the Turing Award in 2018 (the Computer Science equivalent of a Nobel prize) for their work. Today, their revelations have made neural networks the most vibrant area of computer science and have revolutionized fields such as computer vision and natural language processing.

Cancer immunology faced similar obstacles. Treatment with IL-2 cytokine, one of the first immunomodulatory drugs, failed to meet expectations. These outcomes slowed further research, and for decades, cancer immunology wasn’t taken seriously by many cancer biologists. With the effort and intuition of some, however, it was discovered decades later that the concept of boosting the immune system to fight cancer had objective validity. It turned out that we just needed better drug targets and combinations, and eventually, researchers demonstrated that the immune system is the best tool in our fight against cancer.

James P. Allison and Tasuku Honjo, who pioneered the class of cancer immunotherapy drugs known as checkpoint inhibitors, were awarded the Nobel Prize in 2018.

Though widely accepted now, it took decades for the scientific establishment to accept these novel approaches as valid.

Machine learning and immunotherapy have more in common than historical similarities. The beauty of immunotherapy is that it leverages the versatility and flexibility of the immune system to fight different types of cancers. While the first immunotherapies showed results in a few cancers, they were later shown to work in many other cancer types. AI, similarly, utilizes flexible tools to solve a wide range of problems across applications via transfer and multi-task learning. These processes are made possible through access to large-scale data.

Here’s something to remember: The resurgence of neural networks started in 2012 after the AlexNet architecture demonstrated 84.7% accuracy in the ImageNET competition. This level of performance was revolutionary at the time, with the second-best model achieving 73.8% accuracy. The ImageNET dataset, started by Fei-Fei Li, is robust, well-labeled and high-quality. As a result, it has been integral to how far neural networks have brought computer vision today.

Interestingly, similar developments are happening now in biology. Life sciences companies and labs are building large-scale datasets with tens of millions of immune cells labeled consistently to ensure the validity of the underlying data. These datasets are the analogs of ImageNET in biology.

We’re already seeing these large, high-quality datasets giving rise to experimentation at a rate and scale that was impossible before. For example, machine learning is being used to identify immune cell types in different parts of the body and their involvement in various diseases. After identifying patterns, algorithms can “map” or predict different immune trajectories, which can then be used to interpret, for example, why some cancer immunotherapies work on particular cancer types and some don’t. The datasets act as the Google Maps of the immune system.

Mapping patterns of genes, proteins, and cell interactions across diseases allows researchers to understand molecular pathways as the building blocks of disease. The presence or absence of a functional block helps interpret why some cancer immunotherapies work on particular cancer types but not others.

Mapping pathways of genes and proteins across diseases and phenotypes allows researchers to learn how they work together to activate specific pathways and fight multiple diseases. Genes can be part of numerous pathways, and they can cause distinct types of cells to behave differently.

Moreover, different cell types can share similar gene activities, and the same functional pathways can be found in various immune-related disorders. This makes a case for building machine learning models that perform effectively on specific tasks and transfer to other tasks.

Transfer learning works in deep learning models, for example, by taking simple patterns (in images, think of simple lines and curves) learned by early layers of a neural network and leveraging those layers for different problems. In biology, this allows us to transfer knowledge on how specific genes and pathways in one disease or cell type play a role in other contexts.

AI research that addresses the effects of genetic changes (perturbations) on immune cells and their impact on the cells and possible treatments is increasingly common in cancer immunology. This kind of research will enable us to understand these cells more quickly and lead to better drugs and treatments.

With large-scale data fueling further research in immunotherapy and AI, we are confident that more effective drugs to fight cancer will appear soon, thus giving hope to the over 18 million people who are diagnosed with cancer every year.

Source: Tech


Dashworks is a search engine for your company’s sprawling internal knowledge



As a company grows, the amount of important information employees need to keep track of inevitably grows right along with it. And, as your tech stack gets more complicated, that information ends up split up across more places — buried in Slack threads, tucked into Jira tickets, pushed as files on Dropbox, etc.

Dashworks is a startup aiming to be the go-to place for all of that internal knowledge. Part landing page and part search engine, it hooks into dozens of different enterprise services and gives you one hub to find what you need.

On the landing page front, Dashworks is built to be your work laptop’s homepage. It’s got support for broadcasting company wide announcements, building out FAQs, and sharing bookmarks for the things you often need and can never find — your handbooks, your OKRs, your org charts, etc.

More impressive, though, is its cross-tool search. With backgrounds in natural language processing at companies like Facebook and Cresta, co-founders Prasad Kawthekar and Praty Sharma are building a tool that allow you to ask Dashworks questions and have them answered from the knowledge it’s gathered across all of those aforementioned Slack threads, or Jira tickets, or Dropbox files. It’ll give you a search results page of relevant files across the services you’ve hooked in — but if it thinks it knows the answer to your question, it’ll just bubble that answer right to the top of the page, Google Snippets style.

Image Credits: Dashworks

Right now Dashworks can hook into over 30 different popular services, including Airtable, Asana, Confluence, Dropbox, Gmail, Google Drive, Intercom, Jira, Notion, Slack, Salesforce, Trello, and a whole bunch more — with more on the way, prioritized by demand.

Giving another company access to all of those services and the knowledge within might be unsettling — something the Dashworks team seems quite aware of. Kawthekar tells me that their product is SOC-2 certified, that all respective data is wiped from their servers if you choose to disconnect a service, and that, for teams that are equipped to host the tool themselves, they offer a fully on-prem version.

This week Dashworks is announcing that it raised a $4M round led by Point72 ventures, backed by South Park Commons, Combine Fund, Garuda Ventures, GOAT Capital, Unpopular Ventures, and Starling Ventures. Also backing the round is a number of angels, including Twitch co-founder Emmett Shear and Gusto co-founders Josh Reeves and Tomer London. The company was also a part of Y Combinator’s W20 class.

Image Credits: Dashworks

Source: Tech

Continue Reading


Daily Crunch: Google will offer G Suite legacy edition users a ‘no-cost option’



To get a roundup of TechCrunch’s biggest and most important stories delivered to your inbox every day at 3 p.m. PST, subscribe here.

Hello and welcome to Daily Crunch for January 28, 2022! It’s nearly blizzard o’clock where I am, so please enjoy the following newsletter as my final missive before hunkering down. In happier and better news, TechCrunch Early Stage is coming up in just a few months and not only am I hype about it, I’ll hopefully be there IRL. See you soon! – Alex

The TechCrunch Top 3

  • Google invests up to $1B in Airtel: With a $700 million investment and $300 million in “multi-year commercial agreements” with Airtel, and Indian telco, Google has made its second major bet on Indian infra. Recall that Google also put money into Jio, another Indian telco. The deal underscores the importance of the country in the future of technology revenues.
  • What’s ahead for Europe: On the heels of news that European startups had an outsized 2021 when it came to fundraising, TechCrunch explored what’s ahead for the continent. Some expect a slowdown from peak activity, while others anticipate further acceleration. Regardless of which perspective you favor, European venture investment is expected to remain elevated for some time to come.
  • Zapp raises $200M: And speaking of European startups, Zapp, the U.K.-based quick-convenience delivery startup, just raised a massive Series B. The company previously raised $100 million, meaning that this round was big in absolute and comparative terms. As we see some consolidation in the fast-delivery space, this deal caught our eye.


  • Are charter cities the future for African tech growth? TechCrunch’s Tage Kene-Okafor has a great piece up on the site noting that “African cities have the fastest global urban growth rate,” which is leading to overcrowding. Some folks think that “charter cities offer a solution.” Special economic zones of all types have been tried before – will they offer African tech a faster route forward?
  • Personalized learning is hot: Our in-house edtech expert Natasah Mascarenhas has a great piece out today on personalized learning startups – Learnfully, Wayfinder, Empowerly, and others – that are taking the lessons of remote schooling to heart and working to make products that work better for our kids. It’s an encouraging, fascinating story.
  • Rise wants to remake team calendaring: There is no shortage of apps in the market to help individuals and teams work together. But we might not need as many as we have. That’s why Rise is making me think. The team calendaring app just raised a few million, and could replace a few tools that myself and friends use. I wonder if the solution to the Tool Overload of 2022 is tools that do less, intentionally.
  • Canvas wants non-tech folks to be able to squeeze answers from data: Developers are in short supply, so no-code tools that allow folks who don’t sling code to do their own building are blowing up. Similarly, a general dearth of data science talent in the market is creating space for tools like Canvas, which “is going all in with a spreadsheet-like interface for non-technical teams to access the information they need without bothering data teams,” TechCrunch reports.
  • Zigbang buys Samsung IoT business: The IoT promises of yesteryear are coming true, and not. Samsara recently went public on the back of its IoT business. That was a win for the category. That Zigbang, a South Korean proptech startup, is buying Samsung’s IoT unit feels slightly less bullish.
  • Series F-tw? Once upon a time I would have mocked a Series F as indication that the company in question had failed to go public. But that was then. Today Series Fs are not that rare. Indian B2B marketplace Moglix just raised one, which doubled its valuation to $2.6 billion. Tiger co-led the $250 million round.

And if you are looking down the barrel of a blizzard, TechCrunch’s Equity podcast has your downtime covered. Enjoy!

European, North American edtech startups see funding triple in 2021

Image Credits: Bet_Noire (opens in a new window) / Getty Images

Pre-pandemic, VCs were notoriously reluctant to invest in education-related companies. Today, edtech startups are seeing higher average deal sizes, more seed and pre-seed funding from non-VC investors, and an influx of generalists.

According to Rhys Spence, head of research at Brighteye Ventures, funding for edtech startups based in Europe and North America trebled over the last year.

“Exciting companies are spawning across geographies and verticals, and even generalist investors are building conviction that the sector is capable of producing the same kind of outsized returns generated in fintech, healthtech and other sectors,” writes Spence.

(TechCrunch+ is our membership program, which helps founders and startup teams get ahead. You can sign up here.)

Big Tech Inc.

  • Northern Light Venture Capital’s He Huang says the Chinese robotics market is overheated: Per the investor, robotics in China is “riddled with speculation and overvalued companies,” calling the situation a bubble. It’s worth noting that China’s central government is working to retool where its tech investment dollars flow.
  • Robinhood goes down, back up: This morning, in the wake of the company’s lackluster earnings report, TechCrunch dug through why Robinhood’s stock sold off in after-hours, pre-market, and early trading sessions yesterday and today. And then Robinhood turned around and gained ample ground during the rest of the day. It’s a weird market moment, but good news for the U.S. fintech all the same.
  • Google to allow legacy G Suite users to move to free accounts: After angering techies still using the “G Suite legacy free edition” by announcing that it was ending the program and requiring payment, the search giant has decided to ”offer more options to existing users,” TechCrunch reports. Somewhere inside of Google, a business decision just met the market and was flipped on its head. Makes you wonder who is calling the shots over there, and if they previously worked for McKinsey.

TechCrunch Experts

Image Credits: SEAN GLADWELL / Getty Images

TechCrunch wants you to recommend growth marketers who have expertise in SEO, social, content writing and more! If you’re a growth marketer, pass this survey along to your clients; we’d like to hear about why they loved working with you.

Source: Tech

Continue Reading


3 experiments for early-stage founders seeking product-market fit



At Human Ventures, we have a fund for pre-seed and seed-stage investments, a venture studio and an Entrepreneur in Residence (EIR) program.

Through this work, we’ve discovered a lot about how different founders fulfill their journey of customer discovery and product-market fit. One of the largest challenges for pre-seed and seed stage founders is determining where to start: There are a million things to do. What should you do at each stage?

We interviewed three founders from our portfolio, all of whom ran discovery experiments to find their product-market fit at different stages of their company’s development.

Here’s what they had to share:

Pre-MVP/customer discovery phase: Tiny Organics

Tiny Organics is a plant-based baby and toddler food company on a mission to shape childrens’ palates so they’ll choose and love vegetables from their earliest days. The company raised $11 million in their Series A in 2021 and is growing at over 500% annually.

Founders Sofia Laurell and Betsy Fore joined our venture studio as EIRs and went through a six-week discovery sprint. As Sofia explains, they knew they wanted to build something to make parents’ lives easier and threw a lot of initial ideas at the wall from the Finnish baby box 2.0 (Sofia is Finnish) to an easier way to create Instagrammable baby pictures.

They went through multiple exercises to test the viability of new parents’ most pressing and urgent needs:

  • Conduct a “Start with Why” exercise
  • Define the “Jobs to be Done”
  • Create a lean canvas for each (viable) concept
  • Define the user journeys
  • Conduct user surveys using platforms like and 1Q (instant survey tool)
  • Identify and define their customer personas
  • Conduct customer interviews and synthesize them
  • Construct concept prototypes

They also met prospective customers, conducting a focus group of 10-15 moms. When the founders asked them to text them what they were feeding their children along with pictures for a week, they realized the lack of healthy finger foods in the market, thus sparking the idea for Tiny Organics.

Source: Tech

Continue Reading