Category: Uncategorized
「Video of the Week」How deep learning is poised to transform the life sciences and healthcare
Applying deep learning in the life sciences
Deep-learning algorithms have proven to be powerful across many aspects of the life sciences and healthcare. Here are eight examples of how they’re being applied throughout the pipeline:
- Deep-learning-powered lead optimization: Lead optimization is a key problem. Once you’ve gotten your lead, how do you explore a 1080 space to find a better molecule based on your lead? According to a recent publication, a type of deep-learning engine called an “encoder” can encode structures with a discrete library into a continuous gradient space. That allows you to explore a much larger area of compound space than you have in your discrete library. You can then drop novel molecules out of the gradient space, design small compounds and test them to see how they perform, compared with all of the molecules in your existing library.
- Predicting compound activity: One variant of deep learning is known as “one-shot deep learning,” in which you train the algorithm to identify differences, rather than similarities, in data. That requires a lot less training data, which is a critical advantage. A recent paper shows that one-shot deep-learning approaches excel in predicting compound bioactivity based on training with a small set of data. So, if you’re in a new space and don’t know much about the activity of the compounds you’re using, you can still apply these approaches to get a reading of a novel molecule.
- Cell assay imaging analytics: Applying deep learning to cell assay imaging is an obvious path to pursue. All of the papers that have been published in this space so far have shown that deep-learning algorithms can do as well as or better than humans in detecting things like phenotypes. Better yet, they’re more efficient. You still need a human who has to be trained initially and can understand the outcomes. But overall, people are freed up to think about higher-level matters, rather than sitting there looking at images.
- Toxicity prediction: Deep learning has been shown to be significantly more effective than existing methods in predicting the toxicity of any given molecule. Over time, it learns to look for the specific elements or substructures that are causing the toxicity.
- Counterfeit scanning: Deep-learning systems can be trained to detect counterfeit drugs on the web and other sources by examining the package labeling or the pills themselves. They can pick up on little differences between real and counterfeit with a pretty high degree of accuracy. For example, logos or lettering printed on packaging can be slightly off because the presses aren’t exactly the same as the original.
- Electronic health record (EHR) analysis: Deep-learning approaches have been shown to significantly outperform traditional methods in doing things like patient cohort identification, readmissions analysis, clinical trial recruitment and clinical predictive modeling from EHR data stores.
- Language translation: Clinical trial protocol translation is a critical issue. If you provide a protocol in one language and then have someone translate it into another language, how do you know if they have translated all of the important phrases accurately? You can use deep learning to translate back into the original language and figure out how to normalize all those key criteria in your clinical trial.
- Electronic laboratory notebook (ELN) analysis: Many life sciences companies know they have a huge amount of information in their ELNs, but no good way of getting it out. You can have deep-learning systems go in and pull out this information; they can literally read the text in an ELN and figure out what it means.
How deep learning is poised to transform the life sciences and healthcare
「Video of the Week」Robots That Make Coffee
The startup is called Cafe X.
And it has a robotic shop in SF.
Current coffee pricing (2019)
And a similar startup in China – Ratio Master
with its ice coffee priced @ ¥25 in Beijing 合生汇
「Podcast of the Week」a16z Podcast: CEO’s Changing Roles
Technical -> Product -> Sales
How Do E-commerces Generate Revenue? (2)
Following the previous post on Alibaba, here are two other companies similar to Alibaba’s revenue model.
Pinduoduo
GMV (billions, RMB, fiscal 2018) – 471.6 (US$68.6 billion)
revenue (Online marketing services, 2018) – 11,515.58 million
take rate: 2.4418% (of GMV)
revenue (Transaction services, 2018) – 1,604.42 million
take rate: 0.3402% (of GMV)
Revenue total – 13,119,990 k
COGS (Payment processing fees) – 639.29 million RMB
gross margin: 60.14% (of transaction services)
COGS (Costs associated with the operation of our platform) – 2,265.96 million RMB
Overall gross margin: 77.85%
monetization methods
Online marketplace services
Under our current business model, we generate revenues primarily from online marketplace services. Our revenues from online marketplace services include revenues from online marketing services and transaction services.
Online marketing services. We provide online marketing services to allow merchants to bid for keywords that match product listings appearing in search results on our platform and advertising placements such as banners, links and logos. The placement and the price for such placement are determined through an online bidding system.
Transaction services. We charge merchants fees for transaction-related services that we provide to merchants on our platform. As part of our continued efforts to improve user experience, we reward merchants who sell high-quality products and provide superb services with preferential fee rates.
Merchandise sales
From 2015 to the first quarter of 2017, we also operated an online direct sales business under the name of “Pinhaohuo” for certain categories of merchandise such as fresh produce and other perishable products. Under this model, we acquired products from suppliers and sold them directly to buyers. During the time when we operated Pinhaohuo, we also operated our current marketplace model and completed the transition into our current business model in the first quarter of 2017. As a result, our revenues from merchandise sales have decreased substantially from 2016 to 2017, and we no longer generated such revenues after the first quarter of 2017.
Costs of online marketplace services consist primarily of payment processing fees paid to third party online payment platform, costs associated with the operation of the Group’s platform, such as bandwidths and server costs, depreciation and maintenance costs, staff costs and share-based compensation expenses, surcharges and other expenses directly attributable to the online marketplace services. Costs of merchandise sales consist of the same elements as those of online marketplace services, as well as the purchase price of merchandise, shipping and other logistics charges and write-down of inventories.
—
Mogu (IPO filings as of September 30, 2018)
GMV (6 months) – RMB 7.9 billion
revenue (marketing services, 6 month) – RMB 193.05 million
take rate: 2.4437%
revenue (commission revenues, 6 month) – RMB 215.65 million
take rate: 2.7297%
total revenue (6 month) – RMB 489.48 million (6.1959% of GMV)
gross margin: 68.8853%
monetization methods
Marketing services revenues. We generate revenues from marketing services by providing online marketing services, including display-based, search-based and native content-based advertisements and marketing services, to merchants and brand partners.
Commission revenues. We earn commissions from merchants on our platform when transactions are completed and settled. Such commissions are generally determined as an agreed percentage of the value of merchandise sold by merchants. We generally offer two categories of services to our merchants. We typically charge a commission rate of approximately 5% to merchants that choose our entry-level service offerings, which include basic operational support. For merchants who need fuller operational support, we offer incremental services through our prime service offerings, such as support on brand building, fashion influencer matching and deeper data analytics and insights. We generally charge a commission rate of approximately 20% to merchants that opt for our prime offerings.
Other revenues. Our other revenues primarily consist of revenues from financing solutions and other services.
Revenue Recognition
Marketing services revenues
We provide marketing services to merchants and brand partners that help them promote their products in designated areas on our platform directly or via social network platforms over particular periods of time that will then divert users back to our platform. Such services are charged at fixed prices or at prices established through our online auction system. In general, merchants and brand partners need to prepay for the marketing services. Revenue is recognized ratably over the period during which the content is displayed, or when the content or offerings are clicked or viewed by users, or when an underlying transaction is completed by a merchant.
Commission revenues
We operate our online platform for merchants to sell their merchandise to our users and also provide integrated platform-wide services. When the transactions are completed on our platform, we charge merchants commissions at their respective agreed percentage of the amount of merchandise sold by merchants. We identify that arranging for the provision of products by merchants for each successful transaction and the provision of integrated services are separate performance obligations. We apply the practical expedient that allocates the commission revenues for the integrated services to the respective day on which we have the right to invoice. We do not control the underlying merchandise provided by merchants before they are transferred to users, as we are not responsible for fulfilling the promise to provide the merchandise to users and have no inventory risk before the merchandise are transferred to the users or after the control is transferred to the users. In addition, we have no discretion in establishing prices of the merchandise provided by merchants. Commission revenues are recognized on a net basis at the point of users’ acceptance of merchandise.
Commissions are refundable if and when users return the merchandise to merchants and the refund is recognized as variable consideration. We offer refund to merchants based on percentages of estimated commissions of a specific period. We identify the refund as a performance obligation and recognize it as a contract liability. The estimations are reassessed and adjusted at the end of each reporting period.
We also offer volume refund to merchants based on the cumulative sales they generate on our platform during a certain period. Within a certain period, if the total sales generated by a merchant reaches a pre-agreed threshold, the merchant is entitled to a refund of a certain percentage of the commission paid to us. We identify the volume refund as a performance obligation and recognize it at its standalone selling price as a contractual liability. The amount of contractual liabilities involves an estimation of a merchant’s sales amount during a certain period and the related percentage to calculate the volume refund. Such estimation is reassessed and adjusted at the end of each reporting period.
Other revenues
Other revenues primarily consist of the revenues from financing solutions and other services. Financing solutions include loans to users and merchants through factoring arrangements and services to facilitate financial institutions to provide loans to merchants and users on our platform. For financing solutions to merchants and users through factoring arrangements, we record loan receivables when the cash is advanced to the users or merchants, and the service fees are recognized over the term of loans. For services to financial institutions, revenue is recognized when the fund is drawn down by the borrowers or over the financing period on a straight-line basis.
「Video of the Week」Hotel Room Delivery Robots!
These robots can communicate with the elevator, get to the right floor and the right door.
「Podcast of the Week」Jared Diamond: What Can We Learn From Traditional Societies?
The podcast on a book I recently finished – The World Until Yesterday.
「Video of the Week」Smart Restrooms!
And I am seeing smart restrooms at the railway station in Nanjing as well.
「Podcast of the Week」WSJ Instant Message: Google And Facebook Conferences, Face ID Ban In SF
[Announcement] break for 1 week
Was packing up and preparing to travel back to China for 3 month (May – August)
Will be back on May 12.