DeepSeek Unleashes the Revolution of Medical AI

As the dawn of 2023 unfolds, the technological world has witnessed a remarkable emergence: the fully domestically developed generative AI, DeepSeek, has made an astonishing debutOver the past month, DeepSeek has demonstrated exceptional reasoning skills and performance, significantly lowering the application barriers for AI large modelsIt heralds a transformative wave in the medical field and service modalities.

Recent reports indicate that several hospitals have begun to implement the DeepSeek modelNotably, the South China Hospital affiliated with Shenzhen University has taken a pioneering step by locally deploying the Chinese AI large model, DeepSeek-R1. This innovative move marks the beginning of a new chapter in the construction of "AI hospitals".

As Director Wu Song of the South China Hospital states, "AI is an assistive tool, and medical staff are the true core of healthcare." This assertion highlights the collaborative dynamic between human expertise and AI capabilities, emphasizing that technology should enhance, rather than replace, the human element in medical practice.

In terms of practical applications, medical imaging devices, including CT and MRI equipment, have integrated the DeepSeek algorithms to automatically generate structured reportsThis integration has propelled medical device companies such as Wandong Medical, United Imaging, and Mindray to introduce imaging equipment equipped with AI modules, thereby accelerating the embrace of AI technologies by pharmaceutical companies and healthcare institutions.

Currently, the intersection of AI applications in healthcare and clinical practices is deepeningRegarding advancements in AI technology, Xiang Jun, President of United Imaging's X-ray Division, shared insights with reportersHe explained that over two decades ago, AI existed but in forms far removed from today's deep learning systems, incorporating neural networks, statistical methods, and simulations among various technological approaches.

Xiang reflects, "During these past twenty years, my observations of AI gradually became second nature, until the meteoric rise of ChatGPT prompted me to reassess my past cognitive biases

Advertisements

Just as the evolution from primitive to advanced forms is evident in biological history, the human brain’s memory processes remain partially veiled in mysteryHowever, this does not equate to errors in human cognition." This transformative moment shifted his perspective, leading him to embrace AI genuinely.

According to a research report by Huatai Securities, AI technologies are showcasing their potential to reshape how new drugs are developed and how medical services are deliveredBy leveraging intelligent diagnostics, personalized treatments, and drug development, AI significantly enhances the efficiency and accuracy of medical servicesThe ongoing advancements in natural language processing, machine learning, and deep learning suggest a widening application of AI in healthcare, encompassing almost every aspect of medical operations, such as pathological research, drug development, genetic testing, disease screening, assisted diagnosis, imaging analysis, and precision medicine.

We have entered an era of digital transformationToday, AI is reshaping the world at an unprecedented pace, and its impact is particularly evident in the medical domain, where digital transformation is becoming unstoppable.

Xiang further elaborated on the evolution of scientific development, which has traversed four stages: the first being Archimedes’ principle of buoyancy and Galileo's iron ball experiment that represented empirical sciences drawn from objective phenomenaThe second stage, represented by Newton and Maxwell's equations, articulated general rules using precise mathematical language, classifying it as theoretical scienceAlthough this period brought significant advancements, it lacked the brilliance witnessed during the computer technology boom, which drove numerical computations that approached limitless potential, giving birth to numerical sciences applied within fields such as weather forecasting.

Tracing back to the origins of the AI era, it reveals itself as a sage integrating the essence of these preceding stages seamlessly.

The incorporation of AI technology in scientific research has profoundly altered methods and processes within the domain

Advertisements

For instance, in data handling and analysis, AI automates data cleaning and preprocessing, thus elevating data quality and enabling efficient handling and analysis of vast amounts of information to uncover hidden patternsIn imaging and signal processing, AI is prevalent in fields such as medical imaging analysis, astronomical image processing, and biological signal processing, enhancing diagnostic accuracy and efficiencyFurthermore, AI demonstrates immense potential in computational simulations, modeling, and experimental design optimization, accelerating processes like molecular dynamics simulations and optimizing high-throughput screening experiments.

Consequently, AI's emergence signifies that every ordinary individual could potentially become a specialist in a niche field akin to a Newton or MaxwellAI transcends the divide between theoretical science and practical engineering environments, continuously learning and delving into theory while simultaneously engaging in practice to unveil extraordinary functional value.

However, Xiang expresses a concern regarding the increasing prevalence of AI technologies: "As AI technology becomes omnipresent, will we still be able to find true scientists? People might gradually begin to explore theoretical sciences less eagerly." He analyzes the future trajectory of AI, predicting two main directions: firstly, the continuous increase in the number of neurons and optimization of their connections will deepen AI's knowledge base and elevate its intelligent performance; secondly, creating AI with only high school-level knowledge, which, when appropriately trained, can excel in specific specialized fields.

"The concept of 'zero noise technology' effectively marries computational design and deep learning achievements, representing another facet of AI development," Xiang asserts.

He clarifies that 'zero noise technology' differs from noise reduction techniques in opticsIt implies an overall enhancement of the entire imaging chain, optimizing X-ray source technology through meticulous control of tube and high-voltage generator, achieving the best energy curves

Advertisements

Furthermore, modulation of exit spectra ensures optimal outcomes, with calibration against standard reference materials eliminating scatter radiation’s impact on contrastThis array of optimizations encompasses every facet of the imaging chain, in which algorithms play an indispensable role.

Reports indicate that this zero noise technology has been implemented by United Imaging in their Digital Subtraction Angiography (DSA) devices, marking its inaugural entry into clinical applicationThe 'zero noise' DSA allows for noise-free imaging in clinical settings, representing a significant leap in medical imaging technologyThe implementation of this technology not only drastically improves image clarity—including a 57% enhancement in spatial resolution and over a 4-fold increase in signal-to-noise ratio—but also effectively reduces radiation exposure by at least 70%.

Professor Ge Junbo, an academician of the Chinese Academy of Sciences and head of Cardiology at Fudan University’s Zhongshan Hospital, highlighted, "Occasionally, I visit grassroots hospitals and find the image quality is subpar, which adversely affects surgical operations."

This inadequacy often arises from traditional DSA devices failing to clearly display tiny vessels or collateral circulation, adding complexity to surgical proceduresThe success of retrograde Chronic Total Occlusion (CTO) surgeries heavily relies on high-quality imaging support.

Data from CTOCC registered research reflects that approximately 80% of CTO surgeries achieve success in ChinaAdvances in imaging technology, such as coronary CT angiography (CTA), intravascular ultrasound (IVUS), and optical coherence tomography (OCT), are playing increasingly crucial roles in enhancing surgical success ratesMoreover, breakthroughs in domestic medical imaging technologies, such as the zero noise DSA, not only boost image quality but also lower radiation doses, aiding surgeons in executing more precise operations, thereby further confirming success in CTO interventions.

Indeed, the medical industry has extraordinarily high demands for accuracy and safety, posing significant challenges for AI technology.

Looking ahead, how will AI transform clinical practices?

Xiang views the current AI landscape as comparable to specialist physicians, with the future potential for it to evolve into general practitioners

The development trajectories of AI exhibit two prominent trends: on the one hand, after rigorous training, AI showcases profound potential not only in medicine and engineering but also in daily household robotics and appliance management, solving previously insurmountable issues; on the other, it pursues general intelligence by expanding processing capabilities and enhancing computational speed, which institutions like OpenAI and DeepSeek are keenly exploring.

"I find the latter trend particularly thrilling because, with proper training, it could address various industry challenges, resulting in significant advancements in respective sectorsBy integrating with industrial solutions, AI’s development will accumulate experience and progress from quantitative to qualitative transformations, subsequently fueling a vigorous expansion within the AI realm." Xiang asserts that this approach is more impactful and pragmatic than merely pursuing general intelligence, as it directly addresses and resolves real-world issues, continuously refining its capabilities to catalyze the growth of general AI.

Professor Ge Junbo points out that with the continuous advancement of AI technologies, the medical field will witness an in-depth convergence with imaging systems.

For instance, groundbreaking innovations by Konica Minolta indicate promising applications of AI in medical imaging technologiesAt the 7th China International Import Expo, companies like Siemens Healthineers showcased AI-empowered medical imaging devices that are expected to significantly boost the efficiency and accuracy of disease diagnosis and treatmentAdditionally, Kangdelai is introducing AI technologies in its medical device products, further driving innovation within the sectorFollowing imaging procedures, AI will assist doctors in diagnostics, mitigating the occurrences of misjudgments and misdiagnoses while providing optimized treatment strategiesThis transformative wave in AI is ready to take flight.

"I firmly believe that AI will continue to mature and refine itself within catheter labs and contrast dosage management sectors

Soon, Zhongshan Hospital will unveil a novel large medical model, which, after over a year in development and training with vast medical data, demonstrated accuracy comparable to seasoned physicians in numerous evaluations." Ge disclosed that the model exhibits an exceptionally low error rate, akin to the R1 version of ChatGPTZhongshan Hospital plans to offer more detailed data for the model's further learning and is preparing subsequent versions, R2, R3, and beyond.

According to CITIC Construction Investment Securities, significant progress has been made in the application of AI technologies in healthcareFor instance, intelligent diagnostic systems enhance diagnostic accuracy through medical image analysis, personalized treatment plans tailor more precise therapy based on genetic information, and health management tools monitor user health data in real timeThese applications substantially elevate the efficiency and quality of medical servicesCorporations should leverage AI to bolster their product competitiveness and client loyalty, securing their industry positions and competitive edge.

AI is not only fortifying the leading positions of industry giants like United Imaging and Mindray, enhancing the powerful, but it is also opening invaluable opportunities for newcomers and emerging firms to leapfrog and achieve transformative upgradesAI in healthcare is a key element in the international competitive arena for domestic enterprisesObserving the AI strategies of some overseas leaders highlights that aside from independent R&D, mergers, strategic partnerships, and building digital platforms and ecosystems are significant paths to expedite AI advancementsAdditionally, effectively utilizing AI tools like DeepSeek contributes to cost reduction and improved profit margins for enterprises.

Advertisements

Advertisements