prepare a presentation and answer the questions

Professor Hua Zhang, Case Writer Wenying Qian and Research Assistant Shuyang Li of China Europe International Business School prepared this case. It was reviewed and approved before publication by a company designate. Funding for the development of this case was provided by China Europe International Business School and not by the company. CEIBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management.

Copyright © 2020 China Europe International Business School. To order copies or request permission to reproduce materials, call 1-800-545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

CB0093 REV: October 11, 2020

HUA ZHANG

WENYING QIAN

SHUYANG LI

LinkDoc: Commercial Exploration of Healthcare Big Data

At the end of 2014, Zhang Tianze, after five years in the hospital information system (HIS) industry, established LinkDoc Technology. This new enterprise—Zhang’s second startup—would focus on building a healthcare big data platform, which aimed to empower the traditional healthcare industry through big data and artificial intelligence (AI) technologies.

Although innovative, this new endeavor also posed many challenges: How should LinkDoc refine valuable clinical information from massive amounts of raw healthcare data? How should it create commercial value out of healthcare big data?

The complex healthcare industry chain involved many participants, including patients, physicians, pharma (pharmaceutical companies), providers (i.e., healthcare service providers like hospitals, clinics, and laboratories), and payers (i.e., government-run medical insurance and private insurers). Which participants had the strongest willingness to pay for healthcare big data? How should LinkDoc design its products and business model to attract paying customers?

Healthcare big data companies needed significant and sustained investments in R&D. While some companies in the industry had focused merely on R&D from the start—without a clear path to commercialization—and were doomed to failure when encountering a shortfall in external funding, others chose models that yielded quick profits in the areas that made it easier to commercialize big data, but only managed to occupy small corners of the market. Once engaged along one of these paths, none would find it easy to break through and become a household name in the sector. The question for LinkDoc, therefore, was how to strike a balance between short-term survival and long-term ambitious goals.

Company Profile Established at the end of 2014, LinkDoc Technology was an oncology-focused healthcare big data

company. With a headcount of nearly 1,100 in 2018, this startup saw its big data platform cover 60% of oncological disease types, moving ahead of its U.S. counterparts. LinkDoc topped off four financing

D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

2

rounds (see Exhibit 1) with ¥1 billion① in Series D funding, setting a new record for a single round of financing in China’s healthcare AI industry. 1

First Startup in HIS Industry After graduating from the Beijing University of Posts and Telecommunications in 2005, LinkDoc

Founder Zhang Tianze worked successively with Tencent and Alibaba. Coming from a family of physicians, Zhang had great insight into China’s healthcare market. In 2009, China’s total healthcare expenditure reached ¥1,754.192 billion, accounting for only 5.08% of GDP, compared with 7.7% in developed countries. This indicated that China’s healthcare market had plenty of untapped potential. In the same year, China began to vigorously push forward hospital informatization, providing Zhang—a computer professional—with entrepreneurial opportunities.

Zhang started his first business in 2009 by establishing Trustone Net-Technology Co., Ltd., which provided hospitals with HISs. An HIS refers to a platform that provides all departments in a hospital with diagnostic and administrative information through computers and communication equipment. HIS generally comprises an administrative system, a medical management system, a decision support system, and auxiliary systems. Introducing an HIS was a means to bring down the cost of hospital management and improve operating efficiency. As HISs were mainstreamed across hospitals in China, Zhang’s first business got off to a flying start. Trustone served a number of well-known pharmaceutical and medical institutions, with its business maintaining a growth rate of over 100% for several years.2

Repositioning the Market Zhang, however, saw the limitations of the booming HIS market: first, market segments would

soon be saturated. Trustone delivered its HIS services mainly to tertiary class-A hospitals, which limited the size of its potential customer base. Also, it would be hard for Trustone to offer the same services to secondary class-A hospitals or community clinics as it did to tertiary class-A hospitals, due to the differences in their operating models. In addition, this new segment was flooded with highly homogeneous products and services. Secondly, the drive to mainstream HISs only led to a shift from analog in-hospital data to digital, without the capacity to be shared between hospitals. Considering the limited value of data from a single hospital, Zhang attempted to explore a new direction of business development.

Healthcare Data: Healthcare data could broadly be divided into three categories: lifestyle data, health data, and clinical data. The quality of lifestyle and health data could not be guaranteed, adding to the fact that it usually only included a single dimension: as such, it was generally only used for crosschecking. Clinical data, on the other hand, was more reliable and included more parameters, making it more valuable. The downside was that almost all clinical data was owned by individual hospitals, making retrieval of this data difficult, even though this was crucial. Over a period of five years, by introducing HISs into tertiary class-A hospitals, Zhang had succeeded in gathering vast amounts of data, most of it related to hospital management. As a consequence, for the next step in development, Zhang set his eyes on gaining continued access to comprehensive clinical data.

Focus on Oncology: Different diseases posed different challenges and difficulties. Big data and AI technology applied to healthcare held the promise of adding value on three fronts: first, by offering an

① ¥ = CNY = Chinese yuan renminbi; ¥ 1 = approximately US$ 0.1454 on December 31, 2018. D

o N

ot C

op y

or P

os t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

3

online diagnosis platform for common ailments, such as colds and diarrhea; second, by allowing real-time monitoring of chronic diseases, such as diabetes and cardiovascular diseases, through smart wearable devices; and third, by using AI to diagnose and treat serious and rare diseases. Zhang believed that the third category had the highest value and the biggest market potential, especially in the field of oncology. The number of new tumor drugs worldwide surged from 481 in 2006 to 3,286 in 2015.3 According to IMS, tumor drugs would account for 11% of medical expenses by 2020.4

Second Startup: LinkDoc In 2014, healthcare big data startups were mushrooming across China, thanks to rapid

technological advancement and favorable government policy. Focusing on the healthcare big data market, Zhang built his second startup, LinkDoc, at the end of that year.

Excellent Teams

Zhang went to great lengths to headhunt the best international talent to build a team of senior medical experts recruited from big pharma, such as Roche, AstraZeneca, Eli Lilly, Abbott, and Johnson & Johnson; an industrial service team comprising leading experts brought in from McKinsey, KunTuo, and IMS; and a research and data operations team composed of professionals drawn from world-class contract research organizations (CRO) enterprises, such as KunTuo, Covance, and Tigermed, and Baidu, Alibaba, and Tencent (BAT). Over half of these highly skilled staff were engaged in research and data operations. Zhang believed that these teams were capable of building a solid “database” through considerable R&D. With this “database” in place, LinkDoc could operate different business lines for commercialization through data usage.

Characteristics of the Healthcare Industry

In the healthcare industry, the major participants were referred to as the 5Ps, namely, patients, physicians, pharma, providers, and payers. In the consumer goods industry, payers were also beneficiaries, while in the healthcare industry, this was not necessarily the case. Due to a lack of medical expertise, individual patients found it hard to be prescribed the right medication. Therefore, it was the government/insurer that hammered out a list of suitable medications and negotiated prices with pharma. The payer for medications was the government/insurer. As a consequence, pharma was willing to pour funds into clinical research, with a view to getting new drugs onto the list.

Clinical research costs borne by pharma worldwide had been rising year by year. Due to the long cycles and low success rates of drug development, more than 50% of pharma firms around the world turned to professional CROs for clinical research. In 2017, the global CRO market hit US$39.6 billion.① Traditionally, CROs collected clinical data manually at a prohibitively high cost. For example, mining a single patient’s data cost nearly ¥10,000. As a result, traditional CROs conducted most clinical data research only in the pre-marketing stage. In fact, post-marketing clinical data research mattered a great deal to pharma. Masses of post-marketing clinical data, however, still remained untapped. According to the American Society of Clinical Oncology (ASCO), less than 3% of clinical tumor data in the United States was structured for research, while the remaining 97% lay idle in hospitals’ HISs or medical record rooms.5 Any company that could mine this valuable clinical data more efficiently

① According to data from Frost & Sullivan, the global CRO market hit US$39.6 billion in 2017, up 12% year-on-year, and was estimated to reach US$64.6 billion by 2021. CRO penetration was nearly 30% in 2016, and was expected to rise to 46.5% by 2021. D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

4

would be able to tap into the associated blue ocean market: this was LinkDoc’s ambition, although it meant an uphill battle that could not be realized overnight.

As a startup with limited resources and capabilities, the question for LinkDoc was how to build its healthcare big data capabilities.

Building Healthcare Big Data Capabilities LinkDoc chose to break down its goal into a series of small milestones. The first step was to reduce

the cost of data structuring, so that the company could perform post-marketing clinical data research. In early 2015, LinkDoc secured tens of millions of dollars from New Enterprise Associates in the Series A round. Afterwards, instead of rushing to chart a path to commercialization, LinkDoc spent more than a year exploring how it could obtain and structure high-quality clinical tumor data.

Data Mining

Owing to the imbalance in medical resources across China, most high-quality clinical data was in the hands of top tertiary class-A hospitals, which LinkDoc considered its target customers. At the outset, Zhang still chose to reach out to hospital directors to access clinical data as he had done in promoting Trustone—but his advances were met with closed doors. Upon reflection, he realized that LinkDoc had to create value for hospitals in exchange for access to clinical data. But how?

Compared with the hospital directors, department directors/physicians were just a mouse click away from accessing clinical data. They dealt directly with patients and had a deeper understanding of clinical data. More importantly, they needed high-quality clinical data for clinical research, but they were too busy to spare time to refine the raw data for further analysis. Aware of this pain point, LinkDoc decisively turned to these physicians rather than the hospital directors. The company helped physicians structure research-grade clinical data through big data and AI technologies. In return, LinkDoc was able to obtain more comprehensive clinical data. Zhang viewed the process of accessing raw clinical data as like refining crude oil into gasoline. While data mining did not bring LinkDoc any profit, it would lay the foundation for commercialization in the future.

Data Structuring How could raw clinical data on tumors be structured? With HISs applied to hospitals, handwritten

medical records were just imported into Excel spreadsheets. A cancer patient’s medical record, usually over 60 pages long, contained a number of graphs and unstructured data,① without standardized wording, which proved difficult to structure.

Centralized Data Mining: In the traditional data-structuring model, a big data company sent out clerks to hospitals to assist physicians in manual data entry. Then, the data was passed to the backend for manual checking to ensure accuracy through rounds of verifications, which cost a great deal of time and effort. Thanks to its direct link to hospitals’ HISs, LinkDoc was able to send raw data from

① Unstructured data refers to data with an irregular or incomplete structure, and is inconveniently represented by a two-dimensional logical table in the database for lack of a predefined model. Unstructured data includes all formats of office documents, texts, graphs, XML, HTML, statements, images, and audio/video information. Data in the computer information system can be divided into structured data and unstructured data. Considering the diverse formats and standards of unstructured data, it is more difficult to standardize and understand unstructured information than structured information. D

o N

ot C

op y

or P

os t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

5

each hospital to LinkDoc’s backend server for centralized processing. This model eliminated the need to send professionals onsite, leading to substantial cost savings.

Assembly-Line Operations: With raw data available, LinkDoc structured it manually. However, manual data entry was demanding in terms of data identification, and inefficient. It took a qualified clerk two hours on average to collate an electronic medical record. Thus, LinkDoc desensitized medical record data and decomposed it into dozens of units for modular processing, and applied the assembly-line operations model to enhance efficiency. For example, one data-entry clerk was responsible for structuring five or six pages of surgical data in a medical record, and these were treated as independent units (see Exhibit 2). In this way, a new graduate in clinical medicine was capable of processing such a module after 15 to 20 days of training.

“Machine-Assisted Manual Work”: To further reduce manual input errors, LinkDoc devised a double-reading/entry system (DRESS), whereby a medical record was assigned separately to two clerks for data entry (see Exhibit 3). In case of a data discrepancy, the module would be automatically passed on to a third clerk for checking. In addition, LinkDoc developed the Fellow-X intelligent system. After data was imported into the system, digital information would be automatically extracted and entered in a standardized way, and then quality controlled. The combination of machine-assisted and manual work shortened the average time spent processing one medical record from two hours to 17 minutes.

“Automating Manual Tasks”: In order to replace manual work, machines first needed to “understand” the medical records, which were full of non-standardized expressions. For example, an oncology patient’s medical record usually contained thousands of key information points, each of which could be expressed in nearly 100 ways. These non-standardized expressions, too difficult for the machine to understand, required manual annotation by specialists. Therefore, LinkDoc hired clinical experts from top hospitals to do this. With their help, the technical team gradually learned the shorthand commonly used by doctors and established a standardized set of medical terms, so that the machine could learn from accurate, structured data samples. With sufficient data available, LinkDoc’s AI system could understand different expressions of information points in a medical record, and the accuracy of automatic processing improved markedly. Around 95% of the data could be automatically structured through the AI system, while only 5% needed manual verification. LinkDoc shortened the time required to enter a complete medical record from two hours to five minutes, so the cost of data entry dropped from thousands of renminbi to tens of renminbi a piece. By the first half of 2017, LinkDoc’s cancer data platform had collected more than one million pieces of patient data.6

Improving Data through Follow-ups LinkDoc built a follow-up① team of more than 100 specialists with a medical background, in order

to help physicians maintain contact with discharged patients and update feedback data. So as to enhance the efficiency of follow-ups and protect the privacy of patients, LinkDoc came up with research-grade follow-up solutions. Follow-up specialists made calls via the system, without accessing patients’ contact information. Each call was recorded for monitoring. As LinkDoc geared the follow-up model and schedule to different diseases, specialists could not see the complete medical records, but only the questions that were advised to be asked during the 30/90/180-day follow-up

① Follow-up means a hospital keeps in contact with discharged patients for the purposes of medical treatment, scientific research, and teaching, or asks them to undergo a regular re-examination in order to track the effectiveness of the treatment and progression of a disease. In short, follow-up means tracking patients after diagnosis and treatment. D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

6

and those automatically generated by the system based on structured patient data. The follow-up process was similar to an assembly line, with a success rate of more than 90%.

Building a Healthcare Big Data Platform After two years of exploration, LinkDoc accumulated clinical data and post-hospital rehabilitation

data, building a complete structured database. LinkDoc’s data volume increased rapidly from 200,000 pieces in 2015 to 450,000 in 2016, and then to more than one million in 2017. By the end of 2018, LinkDoc had partnered with more than 900 departments in over 500 general hospitals and tertiary class-A special hospitals. The company had processed 625,000+ pages of medical record data per day, structured over 3,800,000 cases, 7 integrated and cleaned more than 10 million pieces of regional medical data, and processed more than 10 million pieces of medical imaging data. LinkDoc saw its big data platform covering approximately 60% of oncological disease types, moving ahead of its U.S. counterparts.

Charting a Path to Commercialization With its database in place, LinkDoc was now in a position to explore various business lines. Its

priority was to cash in on its big data. Among various participants in the healthcare industry, who had the most willingness to pay for healthcare big data? In order to find the most suitable path to commercialization, LinkDoc began to assess the possibilities of value creation for each participant.

Path I: Empowering Physicians

When building data capabilities, LinkDoc targeted physicians to access more data. To facilitate its clinical research, LinkDoc rolled out the HUBBLE system. Through structured big data, HUBBLE brought together medical expertise from various fields to facilitate a tailor-made research approach. HUBBLE also provided users with a statistical methodology and tools, so that they could harness structured data to design research projects, select samples, set variables, and draw statistical charts. HUBBLE also delivered common services, including descriptive statistics, group-by-group comparisons, and survival analyses. 8

Path II: Empowering Pharma

LinkDoc empowered pharma with data insights and consulting, as well as patient recruitment.

Data Insights & Consulting

Big data was able to help pharma understand the market, draw up a marketing plan, and track outcomes. In the past, pharma data sources included market surveys, healthcare authorities, and data companies (e.g., IMS). However, most of this data was sample data with limited coverage and low accuracy, and could not be continuously updated, whereas LinkDoc’s massive data was constantly being updated, and could help pharma enhance marketing efficiency.

LinkDoc developed a special data analysis package to meet the needs of pharma. If a pharmaceutical company was preparing to launch a new drug, the analysis package could shed light on the distribution of relevant patients across regions with a high disease incidence or a high awareness of this therapy. Based on this data, the company could then prioritize selected regions for new drug launches. If similar products already existed on the market, the analysis package could produce a holistic overview of competitors’ market share in these regions. In addition, pharma could D

o N

ot C

op y

or P

os t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

7

use the quarterly/monthly reports from the analysis package to spot trends in key indicators in various provinces, and monitor the implementation of a particular strategy, evaluate its effect, and make adjustments, in case of any problem.

Patient Recruitment

Patient recruitment was the greatest challenge for pharma in clinical trials. Incomplete statistics showed that only 45% of the required patients were successfully enrolled on schedule for clinical trials conducted in China. The main obstacle to patient recruitment was that the selection criteria were too restrictive. Suppose a pharma wanted to develop a drug for thymoma, a cancer of low incidence. In order to perform clinical trials, the company needed to recruit patients who had undergone surgery and postoperative chemotherapy. There were probably only a few hundred patients across China able to meet these criteria. Following traditional models, it was also costly and inefficient to run ads on mainstream media and websites of medical institutions/patient clubs/healthcare portals.

Thanks to its big data platform, LinkDoc was able to recruit suitable patients more efficiently: its database contained structured data about hundreds of thousands of cancer patients. Pharma could quickly select appropriate patients through this big data platform, and engage them in clinical trials with the consent of the physicians. The patients who participated in clinical trials were offered not only generous rewards from pharma, but also new treatment options. Big data-based patient recruitment was a way to help pharma significantly improve the efficiency of R&D and reduce R&D costs.

Path III: Empowering Patients

To empower patients, LinkDoc leveraged its healthcare big data to launch the direct-to-patient (DTP) pharmacy business and online hospital business.

DTP Pharmacy

DTP allowed pharma to license its products to pharmacies directly without intermediaries. Patients with a prescription were able to collect drugs straight from the pharmacy. The DTP business held enormous market potential. In the United States, medication for tumors accounted for 48% of all drugs sold DTP. China’s DTP market was still in its infancy.9 As the Chinese government stepped up healthcare system reforms by separating medical and pharmaceutical services, the DTP market began to thrive.

In 2017, LinkDoc set up smart DTP pharmacies. However, before tapping into the DTP industry, LinkDoc conducted in-depth analyses of patients’ needs at different stages in their treatment, such as disease resistance, healing, and rehabilitation, and then customized an array of DTP services to deliver patient-centered care. LinkDoc’s smart DTP pharmacies provided pharmaceutical services, smart follow-ups, micro-finance, e-prescriptions, medication delivery, medication donations, and patient education. Each pharmacy was staffed with no fewer than four professional pharmacists and experienced clinicians, and was equipped with a UPS cold-chain shipping system. By the end of 2018, LinkDoc had opened over 40 smart DTP pharmacies in four municipalities and 25 provinces, with 1,200 stock-keeping units. LinkDoc was granted exclusive rights to sell certain drugs for serious and chronic ailments, malignant tumors, and rare diseases, serving 20,000 patients in total. 10

Online Hospital

As medical progress prolonged their survival, cancer patients needed both professional treatment in hospital and high-quality rehabilitation services after being discharged. According to authoritative D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

8

research published in 2016 in ASCO Lung Cancer, patients who received timely, regular, and accurate out-of-hospital online rehabilitation guidance had a 26% higher survival rate than those who received offline rehabilitation guidance in an untimely and irregular fashion. Online healthcare was crucial to cancer patients. In April 2018, the Chinese government unveiled the “Internet Plus Healthcare” policy, opening the way for online healthcare services.

Thanks to favorable government policy and financing of ¥1 billion through the Series D round, LinkDoc set up a LinkDoc (Yinchuan) Online Hospital in July 2018, becoming one of the first of a batch of licensed medical institutions. Based on big data platform and other resources, the LinkDoc Online Hospital established patient-centered diagnosis, treatment and rehabilitation systems, which covered services such as cancer knowledge transmission, patient education, online diagnosis, one-stop drug dispensing, intelligent follow-ups, and professional adverse side-effect management. 11 At LinkDoc Online Hospital, tumor patients who sought medical treatment first consulted physicians online, and were then given an e-prescription after diagnosis, and finally bought drugs at LinkDoc’s DTP pharmacy. By the end of October 2018, the LinkDoc Online Hospital had established partnerships with 900+ departments in 400+ hospitals. Providing patients with access to targeted medical services, this platform brought together 11,000+ physicians nationwide, 44% with a deputy senior title or above and 100% with an intermediary title or above 12.

Path IV: Empowering Hospitals

For providers (hospitals), the biggest pain point was the shortage of top physicians that put a squeeze on healthcare services. To address this problem, LinkDoc explored AI-assisted clinical diagnosis and treatment.

Independently developed by LinkDoc, the AI-assisted diagnostic system for pulmonary nodules improved the average accuracy of diagnosis from 60% to 80% and shortened diagnosis times by 25%, compared with average figures from other hospitals in China. Together with Tianjin Science and Technology Commission and Tianjin Chest Hospital, LinkDoc established the first AI-assisted diagnosis and treatment center for lung cancer in the Beijing-Tianjin-Hebei region, which was expected to admit 400,000 local tumor patients, and to leverage the data and algorithm platform to spread the AI-enabled healthcare service all over the country. 13 Moreover, LinkDoc developed a lung metastasis prediction system, which was expected to prolong the survival of 100,000 patients per month in China and reduce the 8,000 misdiagnoses that occurred in China every year.14

Challenges Ahead By 2019, LinkDoc had spent five years cultivating different business lines. Each was at a different

growth stage: which direction pointed to customers with a stronger willingness to pay? How should LinkDoc design and develop its products and services to better help customers and generate payments? With limited resources in hand, how should LinkDoc choose its target market?

The healthcare big data industry was booming, with a huge market potential to unleash. But relevant laws and regulations were not yet in place in China. How should LinkDoc cope with the conflict between the need for profits and reasonable and legal data use?

Zhang demonstrated a strong ability to learn on the job. During his decade-long exploration of different paths to commercialize healthcare big data, he had constantly found new inspiration from industry trends, business lines, and team members. In LinkDoc’s fifth year, while Zhang was gradually making small steps to reach his ultimate goal, he was also aware that a good part of the journey still lay ahead. D

o N

ot C

op y

or P

os t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

9

Exhibit 1: LinkDoc’s Financing

Series Amount Date Investor

Series A Tens of millions of US dollars

January 8, 2015 NEA

Series B Ten million US dollars April 4, 2016 Cenova Capital; NEA; CBC; Ally Bridge

Series C Tens of millions of US dollars

May 31, 2017 Long Hill Capital; Temasek

Series D ¥1 billion July 4, 2018 China Investment Corporation; Long Hill Capital; NEA; CBC; Ally Bridge

Source: Provided by LinkDoc.

D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

10

Exhibit 2: Example of LinkDoc’s Structuring of Data in Oncological Patient’s Medical Record

Source: Provided by LinkDoc.

D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

11

Exhibit 3: LinkDoc’s DRESS Engine

Note: QC means quality control.

Source: Provided by LinkDoc.

D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860

LinkDoc: Commercial Exploration of Healthcare Big Data CB0093

12

Endnotes

1 “LinkDoc Raised ¥1 Billion in Series D Round Led by State-Level Investors, Setting a New Record for a Single Round of Financing in China’s Medical AI Industry,” AI Finance & Economics, July 4, 2018, accessed August 13, 2018, http://baijiahao.baidu.com/s?id=1605060334941941289&wfr=spider&for=pc.

2 "Zhang Tianze," cyzone, accessed August 18, 2020, https://data.cyzone.cn/content/dbase/founder?content_id=424811. 3 Mao Yanyan and Gao Liubin, “Report on Global Development of Antitumor Drugs (2016),” Science & Technology Review 34,

no. 11 (2016), accessed August 13, 2018, https://wenku.baidu.com/view/5fefe109876fb84ae45c3b3567ec102de2bddfe9.html.

4 Yue Yuan and Wang Jue, “China’s Tumor Treatment Market Will Enter Golden Decade (I),” yyjjb.com, December 12, 2016, accessed August 13, 2018, www.yyjjb.com/html/2016-12/12/content_245327.htm.

5 Li Haoyuan, “LinkDoc CTO Luo Ligang: AI Makes Medical Research Simpler,” IT Manager World, June 7, 2017, accessed January 10, 2019, www.sohu.com/a/146881935_740071.

6 Ibid.

7 “LinkDoc, a New Unicorn, Is Capable of Monetizing Medical Record Data,” China Pharmaceutical Innovation and Research Development Association, December 21, 2018, accessed March 10, 2019, www.phirda.com/artilce_18994.html?cId=1.

8 “Zhang Tianze: Value of Technology Is to Help Physicians and Benefit Patients,” Sohu, June 20, 2017, accessed January 10, 2019, https://www.sohu.com/a/150536393_740071.

9 Zhao Haoran, Changjiang Securities, Special Report on DTP: LinkDoc Set up DTP to Sell Prescription Drugs (2017), accessed August 13, 2018, http://muchong.com/html/201708/11593563.html.

10 Liu Kuang, “LinkDoc Harnessed Big Data to Set up DTP Pharmacies,” OFweek, December 14, 2018, accessed March 20, 2019, http://mp.ofweek.com/medical/a645673023666.

11 “In Partnership with Yinchuan Health and Family Planning Commission, LinkDoc Online Hospital Harnessed Big Data to Deliver Internet Plus Healthcare Service,” Sina, July 29, 2018, accessed March 30, 2019, http://k.sina.com.cn/article_5334569296_13df7115002000abal.html.

12 “LinkDoc Online Hospital Put Prescription Process Online to Provide Closed-Loop Healthcare Service,” Qimay, October 26, 2018, accessed March 20, 2019, www.sohu.com/a/271442623_656199.

13 Ning Xinyan, “LinkDoc Unlocks Value in Medical Big Data,” Entrepreneur Country 9 (2018), accessed January 10, 2019, www.fx361.com/page/2018/0917/4231544.shtml.

14 Ibid.

D o

N ot

C op

y or

P os

t

This document is authorized for educator review use only by Sun Tam, Framingham State University until Jan 2022. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860