Market Scan of AI in Healthcare Development (Part 2): Exploring the Infinite Possibilities of Future Health Management Beyond Hospitals
-
The healthcare industry has an extensive value chain. The previous article, Market Scan of AI in Healthcare Development (Part 1), discussed the value and prospects of AI-assisted diagnosis in hospital settings.
This article focuses on scenarios outside hospitals, where AI empowers health management. It explores how AI technology connects doctors and patients after they leave the hospital and helps patients understand their current condition and predict disease progression.
Table of Contents
China's healthcare industry is vast, reaching 6.2 trillion yuan in 2017, with an annual growth rate exceeding 10% in recent years.
The healthcare industry includes pharmaceuticals, elderly care, medical services, health supplements, and health management services. Health management accounts for only 2.6% of the total healthcare market but still represents an annual industry scale of nearly 180 billion yuan. Additionally, health management is closely intertwined with medical services, pharmaceuticals, and elderly care.
Due to the current shortage of medical resources, doctors struggle to monitor patients after they leave the clinic, and patients lack timely updates on their condition. Both doctors and patients urgently need health management solutions, particularly in chronic disease care.
With technological advancements, portable medical-grade sensors now provide convenient and accurate health data collection. The low-latency edge computing framework of smart portable sensors enables near real-time feedback, allowing doctors and patients to promptly respond to changes in health status.
Based on real-world patient data and medical big data, more disease dynamics and population physiological-pathological models are being developed, facilitating personalized medical solutions and making precision medicine tailored to individual needs a reality.
Chronic Diseases: A Growing Concern
Beyond the impact of an aging population, some chronic diseases are increasingly affecting younger demographics.
A 2019 report by Health News and Dingxiangyuan revealed that young people face various health issues, particularly in digestion and sleep.
Millennials (born in the 1980s) are increasingly interested in health and fitness knowledge. The top three health topics of interest are diet and nutrition, exercise, and home medication—all closely related to out-of-hospital health management.
Chronic diseases are diverse, and rehabilitation, medication refills, and other behaviors often occur outside hospitals. Daily habits and environments significantly influence disease progression. Using home medical devices to monitor conditions is a common practice among chronic disease patients.
Take diabetes as an example: in 2017, China had 114 million diabetic patients, and due to lifestyle and environmental factors, this number is projected to reach 154 million by 2045. In 2017, direct medical expenses for diabetes accounted for 13% of national healthcare spending, totaling 173.4 billion yuan.
From a macro perspective, the WHO's Innovative Care for Chronic Conditions Framework (ICCC) emphasizes comprehensive, time-sensitive, and patient-centered care, including self-management skills.
Given China's severe shortage of doctors, patients in community and home settings often lack direct medical guidance. From a micro perspective, doctors need real-time updates on patients' physical and mental conditions outside hospitals to provide personalized medication and lifestyle recommendations based on medical history and individual differences.
AI's Role in Chronic Disease Management
AI-powered chronic disease management, backed by medical expertise, plays a crucial role here.
By leveraging wearable sensors to monitor vital signs, medical big data for patient history, and AI-driven physiological or metabolic models from real-world case studies, doctors can remotely manage conditions. If abnormal data exceeds thresholds, they can intervene promptly, adjust treatment plans, and achieve dynamic, end-to-end chronic disease supervision.
Patients can also enhance self-management through AI-driven health feedback. This dual approach—doctor-led and patient-engaged—establishes systematic, personalized chronic disease management, optimizes healthcare resource utilization, and ultimately improves patient health outcomes.
AI's Expanding Value in Health Management
AI's application in health management is increasingly demonstrating its value.
A recent study in Nature Medicine [1] found that real-world data-based models for diabetic kidney disease outperformed clinical trial-based models. This suggests that integrating more real-world patient data improves AI's predictive accuracy for chronic diseases like diabetes.
In 2018, Andrew Ng's team used logistic regression to show that ARR predictions often correlate with cardiovascular risk [2]. Their new X-learner algorithm revealed that individual treatment effects frequently deviate from baseline risk. This finding highlights the superiority of personalized hypertension treatment over one-size-fits-all approaches, offering new perspectives for existing therapies—precisely the outcome Ng's team aimed for.
The focus of health management lies outside hospitals, and the challenge lies in data monitoring.
In September 2014, TIME magazine's cover story Never Offline envisioned how wearable technology would transform lives. Just a few years later, medical-grade wearables have diversified monitoring methods and improved accuracy for patients and doctors outside clinical settings.
Medical-Grade Wearables: Leading the Charge
The most notable medical wearable is arguably Apple Watch Series 4.
Launched in 2018 with FDA clearance, its single-lead ECG and proprietary algorithm enable high-precision heart rate monitoring, alerting users to atrial fibrillation and improving early stroke detection.
Unfortunately, Apple Watch Series 4 lacks approval from China's National Medical Products Administration (NMPA, formerly CFDA), preventing ECG functionality in China.
Beyond the Apple Watch, Apple holds a patent for embedded sensors in AirPods to track temperature, perspiration, heart rate, and other metrics. However, no new AirPods were announced at Apple's September 10 product launch.
Apple Watch Series 5 upgraded its heart rate algorithm and added noise detection and menstrual cycle tracking.
Global tech giants like Apple, Samsung, and Google are investing heavily in medical wearables for health management. Chinese companies are keeping pace: Huami, a top global wearable vendor, released the Amazfit Health Watch in June 2019—China's first wearable with ECG monitoring.
Huami also offers an NMPA-cleared ECG smartband. Leveraging wearable data, medical expertise, and AI, Huami explores applications in cardiac diagnostics, sleep monitoring, fitness, and biometric ID.
In March 2019, Huawei added Class II medical device sales to its business scope, sparking speculation. Collaborating with 301 Hospital, Huawei developed a heart health app using smartwear heart rate sensors and AI algorithms for arrhythmia screening, personalized guidance, appointment booking, and integrated management.
At its September 20 launch, Huawei Watch GT2 added blood oxygen monitoring. However, like most wearables in China, it lacks NMPA approval, meaning its data is for reference only—a common industry hurdle.
Beyond smartwatches, digital blood pressure monitors, glucose meters, and pulse oximeters with AI diagnostics (some NMPA-approved) are already on the market.
In recent years, tattoo-like electronic skin based on biosensors, capable of monitoring human signals such as respiration, heart rate, and vocalization, has made significant breakthroughs in laboratories at Tsinghua University, MIT Media Lab, Stanford University, and others. For example, Professor Zhenan Bao from Stanford University recently developed the wearable wireless skin sensor BodyNet, which requires no battery. It operates similarly to an ID card, using radio frequency identification (RFID) technology to absorb energy from clothing receivers to power the sensor, then reads data from the skin and sends it back to the receiver.
BodyNet will be used in medical applications, such as monitoring the sleep quality of patients with sleep disorders, tracking their respiration, heartbeat, and muscle activity at night, or observing the cardiac performance of heart disease patients in their daily lives over the long term.
The future seems promising. Leveraging the latest sensor technology, internet technology, and AI, a comprehensive health management system can be formed, spanning from out-of-hospital data monitoring, auxiliary diagnosis, and patient reminders to quick appointment scheduling, telemedicine, and pharmaceutical e-commerce, integrating both online and offline processes. Both internet companies and traditional medical institutions have opportunities to participate.
For end-users (C-side), the focus is on selling hardware that detects vital signs and corresponding value-added services, enabling ordinary people to observe accurately quantifiable changes in their health metrics, thereby raising awareness of health management.
At the same time, these comprehensive quantitative data serve as crucial evidence for auxiliary diagnosis and the starting point for health management. For business clients (B-side), the appeal lies in data analysis platforms, cloud management platforms, and AI model platforms. By continuously empowering medical institutions with technological capabilities, patient health management efficiency can be improved, forming a multi-faceted closed loop involving in-hospital and out-of-hospital care, patients, doctors, and platforms.
Under the premise of ensuring privacy and data security, the more detailed and comprehensive the data records, the more likely it is to develop clear and accurate health management plans, making personalized precision medicine a reality.
Currently, companies involved in health management have ambitions beyond this. Starting with health management, they aim to collect vast amounts of patient or user data across multiple scenarios, build digital platforms for medical institutions, and leverage C-side user (patient) scale and B-side data analysis capabilities to enter the larger and more profitable market of auxiliary diagnosis.
[1] Stefan Ravizza. (2019). Predicting the early risk of chronic kidney disease in patients with diabetes using real-world data – https://www.nature.com/articles/s41591-018-0239-8
[2] Tony Duan. (2019). Clinical Value of Predicting Individual Treatment Effects for Intensive Blood Pressure Therapy – https://www.ahajournals.org/doi/10.1161/CIRCOUTCOMES.118.005010
[3] Bao Group Official Website – http://baogroup.stanford.edu/
Additional Resources:
[1] NMPA-Certified Medical Device Information Query: http://app1.sfda.gov.cn/datasearchcnda/face3/dir.html?type=ylqx