All Chinese internet giants are pouring huge sums of money into the same thing.

From ByteDance to Alibaba, then to Tencent.

Overnight, the leading tech giants in China's internet industry all seem to have made the same decision: pour in capital, resources and technical strength to fully explore the "AI + Healthcare" track.

How Far Are We From "Doubao Pharmaceuticals"

After people got access to the "Doubao smartphone", we may also get to take "medicines developed by Doubao" in the near future.

In the past September, Reuters cited sources familiar with the matter saying that Anew Labs, an AI pharmaceutical company spun off internally from ByteDance, has just completed a $290 million financing plan.

During this round, in addition to HSG, IDG Capital and GL Ventures leading the investment, well-known industry players including 5Y Capital, Chunhua Ventures, Boyu Capital and others are also listed in the circulated investor list.

Against this backdrop, Anew Labs' valuation once soared to $1.5 billion.

When you open its official website, the first thing that catches your eye besides the slogan "Fight Disease Advance Human Health" is the most notable platform matrix that covers everything from protein structure prediction to supporting drug discovery decision-making.

As a lab backed by ByteDance, the core reason why Anew Labs is favored by investors from all sides lies in the possibility of ByteDance's algorithm capability spillover. In the digital space, ByteDance's recommendation algorithm can instantly analyze and predict users' behaviors and preferences, and accurately locate the video that is "most likely to retain users' attention" from the vast content library.

Apply this logic to the pharmaceutical industry.

We will find that the essence of drug molecular design is also a search game that looks for the "optimal solution" in the chemical space.

What are the pain points of traditional pharmaceutical R&D? High cost and long cycle. For this reason, the industry has the double "10" development law of "1 billion US dollars + 10 years of cycle".

What Anew Labs aims to do is to use algorithms and AI to improve the efficiency of molecular screening and structure prediction, so as to greatly cut time cost and accurately lock in candidate molecules that are most likely to become "active ingredients".

This logical reconstruction at the efficiency level is probably the underlying reason why the capital market is willing to give this young company a high valuation.

ByteDance has also found a new AI monetization path in this game, even though it has just started.

Alibaba's Medical Imaging AI is Published in Top International Academic Journal

In DAMO Academy, Alibaba has also set its sights on the analytical efficiency of AI, which far outperforms human capabilities.

On September 18, DAMO RADAR, a general medical imaging AI model jointly developed by Alibaba DAMO Academy and the First Affiliated Hospital, Zhejiang University School of Medicine, was successfully published in the top international journal Science.

It is different from those AI models trained for specific targets, which adopt the "one disease, one model" system.

The most prominent feature of DAMO RADAR is that it realizes the precise identification of 146 diseases in 18 abdominal anatomical organs with a single model for the first time.

"The 'one disease, one model' system relies heavily on manual annotation, which costs a lot. Usually, it takes 2 years to train a model for one specific disease, and the trained model cannot be generalized to other diseases."

Representatives from DAMO Academy said that DAMO RADAR is like a "navigation software" for doctors. Its value is not to replace doctors to make final diagnoses, but to maximize the sensitivity of detecting pathogenic abnormalities and improve the work efficiency of all doctors.

"The core value of disease-specific models is high precision, while the core value of general models is wide coverage."

As a new choice with completely different technical solutions and value ranges, in terms of data, through human-machine collaborative testing, researchers found that influenced by AI prompts, radiologists' sensitivity in disease identification increased by about 10%, and the average (single case) image reading time decreased by 30.7%. In some test sessions, junior doctors assisted by AI even outperformed senior doctors in terms of sensitivity.

The improvement of sensitivity means that the probability of abnormal areas and tissues being detected rises significantly. Regrettable incidents such as missed diagnosis will become increasingly rare.

The improvement of image reading efficiency means that doctors can help more patients within the same working hours, so that more patients can get medical help in time.

Tencent Cloud Focuses on the "Logistics Support" Sector

If ByteDance and Alibaba's medical entry solutions are more or less To B for C, that is, to further improve the treatment experience of individual patients by empowering medical capabilities.

Then Tencent's vision is set on the more "hardcore" field of B x B.

In August, Tencent Cloud Intelligence and Mindray Medical, a long-established domestic medical device manufacturer, jointly launched the world's first large model for medical technology services, to provide more flexible technical support services for engineers in the equipment departments of hospitals and other relevant personnel.

Industry insiders pointed out that with the increase of the number of various devices and the types of in-hospital equipment, the daily pressure faced by operation and maintenance personnel has risen to a certain extent. To avoid the situation where "when an emergency occurs, on-duty personnel have to manually go through thick operation manuals and wait on site for remote support from engineers".

A redundant solution that can respond quickly 7*24 hours regardless of geographical location has become the focus of technical personnel from Tencent Cloud and other institutions.

Nowadays, with the help of AI, no matter it is late at night or not, no matter where you are, on-duty personnel only need to describe the problem to the AI system in natural language, and the model will immediately give corresponding troubleshooting steps and precautions.

(In terms of data, in the professional evaluation test, the above model has a 99% accuracy rate in answering key questions; its response speed is also sufficient to support second-level response for millions of service consultations per year.)

What is more exciting is that the person in charge of Mindray Medical said that the value of the model is far more than that.

With the continuous convergence of global device service data, AI will further understand the law of equipment failure, the distribution of user demands and the changes of service scenarios, and predictive maintenance, management decision support and other clinical application scenarios are gradually becoming a reality.

From ByteDance to Alibaba, then to Tencent, this diversified entry path from "assisting R&D" to "treating patients and saving lives" and then to "equipment guarantee" also reflects the common ambition of China's leading internet giants in the medical track: they are no longer satisfied with just being traffic distributors or tool providers, but try to deeply reconstruct the underlying efficiency logic of the medical industry through AI technology.

Admittedly, rational thinking is indispensable amid the upsurge.

The particularity of the medical industry determines that it cannot carry out trial and error and evolution through simple "rapid iteration and small-step trial" like internet products. Every algorithm update is tied to the weight of life and ethical considerations.

But in any case, this "medical AI arms race" led by leading tech giants has shown us a more accurate, more efficient medical scenario that benefits ordinary people more. This kind of "product competition" is probably exactly what the public is happy to see.

References:

Tianyancha

IT Home: Reports Say ByteDance Completed $290 Million Financing to Spin Off AI Pharmaceutical Business Anew Labs

Titan Media APP, AI Capital View: AI Pharmaceutical Unicorn Anew Goes Independent, ByteDance Pushes Forward the "Most Capital-consuming Slow Business"

CYOL.com, The Paper: The First Expert-level General Medical Imaging AI is Published in Science

Sina Finance, Securities Times: The First Large Model for Technical Services in the Medical Industry is Launched

This article is from WeChat Official Account "mawen011" (ID: hlw0823), written by Internet Stories, authorized for release by 36Kr.