Advertisements
As we step into a new year, a remarkable topic has quietly emerged within the tech community—DeepSeek, a homegrown generative AI, is making wavesIn just a month, DeepSeek has significantly reduced the complexities surrounding the application of AI large models, hinting at revolutionary changes within medical scenarios and service models.
Several hospitals have already integrated the DeepSeek large model into their systemsNotably, Shenzhen University’s South China Hospital took a pioneering step by locally deploying the domestic AI large model, DeepSeek-R1, marking the beginning of a new chapter in the construction of “AI hospitals.”
Wu Song, the hospital director of Shenzhen University’s South China Hospital, emphasized, “AI is a supportive tool; healthcare personnel are the true core of medical work.” This statement underlines the importance of human intuition and expertise in an industry often intertwined with technology.
Meanwhile, AI is proving transformative in various application scenariosMedical imaging devices such as CT and MRI scans are now embedding DeepSeek algorithms that automatically generate structured reportsThis advancement has propelled medical device manufacturers, including Wandong Medical, United Imaging, and Mindray, to launch imaging equipment equipped with AI modules, fast-tracking the adaptation of AI by pharmaceutical companies and medical institutions.
The integration of AI into medical practice is deepening rapidlyXiang Jun, president of the X-ray division at United Imaging, commented on the evolution of AI technology. “AI existed over twenty years ago, but it was not in the form we see today with deep learningIt embodied a variety of technological paths including neural networks, statistical methods, and simulation,” he stated.
“Over the past two decades, my observation of AI has gradually become normal as we moved towards the rise of ChatGPT, which made me reflect on my previous cognitive biases
Advertisements
Just as biological evolution moves from lower to higher forms, the complexities of the human brain’s memory mechanisms remain elusive, which should not be seen as proof of human cognitive error,” he continued, marking a personal enlightenment towards accepting the potential of AI.
According to a report from Huatai Securities, AI technology is showcasing its potential to reshape the pharmaceutical development models and healthcare deliveryThrough applications such as intelligent diagnostics, personalized treatments, and drug development, healthcare service efficiency and accuracy have been significantly enhancedThe ongoing evolution of natural language processing, machine learning, and deep learning indicates a vast and expanding role for AI across virtually all healthcare segments—spanning pathology research, drug development, genetic testing, disease screening, assistance in diagnostics, image analysis, and precision medicine.
We have entered an era of digital transformationAI technology is rapidly altering the landscape of the world, with the medical field exemplifying an unstoppable shift towards intelligentization.
Xiang Jun notes that the history of scientific development has involved four distinct phases: The first phase includes Archimedes’ principle of buoyancy and Galileo’s iron ball experiments, which summarized laws from objective phenomena and belong to the realm of empirical scienceThe next phase, characterized by Newtonian physics and Maxwell’s equations, expresses general laws using precise mathematical languageHowever, this stage appears somewhat lackluster in its allureThe advent of computational technology has given rise to computational science, allowing for numerical computations that, while less precise than symbolic ones, possess near-unlimited potential in applications like weather forecasting.
Tracing back to the genesis of the AI era reveals it as a synthesis of the prior three phases, cleverly amalgamating their essences.
AI technology’s application in scientific research is profoundly changing the methodologies and processes of inquiry
Advertisements
In data handling and analysis, AI enhances data quality through automated cleansing and preprocessing, efficiently processing vast quantities of information to uncover hidden patterns and lawsIn the domain of image and signal processing, AI is widely applied in medical imaging analytics, astronomical image processing, and biological signal processing to boost diagnostic precision and efficiencyMoreover, AI demonstrates immense potential in computational simulation, modeling, experiment design, and optimization, accelerating molecular dynamics simulations and refining high-throughput screening protocols.
The advent of AI heralds a newfound opportunity for every individual to potentially emerge as a specialist in a niche field much like Newton or MaxwellBy bridging theoretical knowledge with practical applications, AI showcases its extraordinary utility.
“I harbor a concern regarding the widespread adoption of AI technology: will we still hold on to the essence of what it means to be a true scientist?” Xiang Jun wondersHe believes that the future trajectory of AI will revolve around two major directions: One is the continuous increase in the number of neurons and the optimization of their connections to amplify AI knowledge; the other is the creation of AI with a high school-level understanding that, with appropriate training, performs excellently in specific niches.
“Zero-noise technology essentially combines computational design principles with deep learning outcomes, representing another facet of AI advancementI firmly believe that with China’s solid industrial foundation, we can leverage AI as a universal key to unlock numerous challenges within the industrial sector,” Xiang Jun asserts.
He further differentiates zero-noise technology from traditional optical denoising techniquesZero-noise technology optimizes the entire imaging chain through enhancements in X-ray source technology, including meticulous control of the tube and power generator to achieve optimal energy curves; modulation of emitted spectra to attain the best spectrum; and calibration of detectors using standard reference materials to filter the impact of scattered signals on contrast
Advertisements
This series of enhancements is crucial as algorithms play a pivotal role.
Reports indicate that this zero-noise technology has already been applied by United Imaging in DSA (Digital Subtraction Angiography) devices and is being used clinically for the first timeZero-noise DSA signifies a significant breakthrough in medical imaging technology, as it vastly improves image clarity, enhancing spatial resolution by 57% and boosting the signal-to-noise ratio by over quadruple, while successfully reducing radiation doses by at least 70%.
Professor Ge Junbo, an academician of the Chinese Academy of Sciences and director of the cardiology department at Zhongshan Hospital affiliated with Fudan University, mentioned, “Sometimes when I visit grassroots hospitals, the image quality of equipment is suboptimal, affecting surgical procedures.”
This predicament arises as traditional DSA devices often struggle to clearly visualize small vessels or collateral circulation, thereby complicating surgical operationsThe success of reverse CTO surgeries relies heavily on high-quality imaging support.
According to data registered by CTOCC, the success rate of CTO procedures in China is approximately 80%. Advances in imaging technologies like coronary CT angiography (CTA), intravascular ultrasound (IVUS), and optical coherence tomography (OCT) increasingly facilitate improved surgery success ratesFurthermore, breakthroughs in domestic medical imaging technologies, such as zero-noise DSA, not only elevate image quality but also diminish radiation exposure, assisting doctors in executing more precise operations, thus enhancing the likelihood of successful CTO procedures.
The medical industry has a feverent requirement for precision and safety, placing enormous demands on AI technology.
The question arises: how will AI transform the future of clinical practices?
Xiang Jun contemplates that current AI is similar to specialized physicians but may evolve into the role of general practitioners in the future
Two prominent trends are surfacing in AI development: on one hand, AI is showcasing immense potential in specialized fields like medicine and engineering while also tackling challenges in home robotics and appliances; on the other hand, there’s a pursuit of general intelligence—expanding processing capabilities and increasing computational speed—which institutions like OpenAI and DeepSeek are exploring.
“I am particularly enthusiastic about the second direction, as it will potentially solve various industry problems with adequate training, significantly pushing the boundaries of multiple sectorsThe interaction between industries to address practical problems will yield cumulative effects, transitioning from quantitative changes to qualitative shifts, and subsequently leading to robust growth in the AI sector,” Xiang Jun believesHe argues that this path, rather than the pursuit of generalized intelligence, is more pragmatic and rapid, as it directly addresses and resolves real-world issues, thereby fostering the overall development of general AI.
Professor Ge Junbo noted that with the ongoing advancements in AI technology, we are likely to witness deeper integration between AI and imaging systems in the healthcare sector.
For example, Konica Minolta's innovative breakthroughs indicate promising applications of AI in medical imaging technologyAt the seventh China International Import Expo, companies like Siemens Healthineers showcased AI-enabled medical imaging equipment that is set to enhance diagnosis and treatment efficiency significantlyKangdelai has also integrated artificial intelligence into its medical devices, further driving innovation within the healthcare sectorPost-imaging, AI will assist doctors in diagnostics, reducing instances of misjudgment and misdiagnosis, and providing optimized treatment plans for patientsThis transformation is on the cusp of realization.
“I firmly believe that AI will become increasingly mature and refined in areas like cath lab environments and contrast dose management
Soon, Zhongshan Hospital will launch a new medical large model, developed over more than a year and trained with extensive medical data, whose accuracy is validated to match that of seasoned doctors in multiple assessments,” revealed Ge JunboThis model boasts an extremely low error rate, akin to the ChatGPT R1 versionZhongshan Hospital will provide more detailed data for the model's further learning, anticipating subsequent versions like R2, R3, and beyond.
According to China Citic Securities, the application of AI technology in the medical field has made remarkable strides, with intelligent diagnostic systems enhancing diagnostic accuracy through medical imaging analysis, personalized treatment plans based on patient genetic information, and health management tools for real-time user health data monitoringCollectively, these applications are significantly amplifying the efficiency and standards of healthcare servicesCompanies are expected to leverage AI to elevate product competitiveness and customer loyalty while solidifying their industry status and advantages.
AI not only consolidates the leading positions of industry giants like United Imaging and Mindray but also opens up valuable opportunities for emerging enterprises to leapfrog and transformAdditionally, AI in healthcare is a critical component of domestic enterprises’ global competitive strategiesObserving the AI frameworks from leading overseas companies, beyond independent R&D, mergers, strategic collaborations, and establishment of digital ecosystems are pivotal paths to accelerate AI developmentAdditionally, effectively employing tools like DeepSeek will aid firms in reducing costs, boosting efficiency, and enhancing profit margins.
“Globally, no other nation aside from China can integrate AI with industry so seamlessly to tackle real-world industrial challengesThis process heralds a profound transition from quantitative to qualitative leapsAs AI applications proliferate, this will further fuel the advancement of general AI in terms of increasing neuron numbers and computational speeds,” Xiang Jun highlighted
Advertisements
Advertisements
Leave a Comment