DeepSeek and the Democratisation of LLM's

Professor Shafi Ahmed

Surgeon | Futurist | Innovator | Entrepreneur | Humanitarian | International Keynote Speaker “The future of healthcare lies in our ability to harness the power of AI—not as a tool, but as a partner in healing. From diagnostics to patient care, AI is redefining what’s possible, and AI models like ChatGPT and now the new kids on the block, DeepSeek and Alibaba’s Qwen are leading the charge.” – Professor Shafi Ahmed

Welcome to the latest edition of AI Horizons, where I dissect the synapses of innovation shaping our medical future. 

I have been travelling for the last month to various global Healthtech events, and so apologies for the delay in writing this newsletter. 

Arab Health in Dubai was a melting pot and possibly the largest gathering of health companies worldwide, with over 150,000 attendees. The Global AI Summit in Cannes followed on from the AI summit in Paris, where health regulation and drug discovery took centre stage. I will cover the exciting topic of AI and Drug discovery in a newsletter next month.

Today, we embark on a journey through the labyrinth of newer large language models (LLMs)—systems redefining the core of healthcare. There has been a lot of discussion and interest in the launch of DeepSeek AI, which is challenging the existing players in the market. But what does that mean for healthcare? anuary also saw the launch of Qwen from Alibaba.

Deepseek

The story of DeepSeek is fascinating and has questioned Silicon Valley’s model of developing LLMS at a high price point and which had not been questioned previously.

DeepSeek, an AI development firm based in Hangzhou, China, was founded in May 2023 by entrepreneur Liang Wenfeng. Prior to establishing DeepSeek, Liang co-founded the quantitative hedge fund High-Flyer in 2016, which managed over $10 billion in assets by 2019. Leveraging his experience in finance and technology, Liang aimed to create an independent AI research lab under the High-Flyer umbrella, focusing on developing advanced AI models. 

DeepSeek gained significant attention in January 2025 with the release of its AI chatbot, DeepSeek-R1, a model boasting 685 billion parameters. What's truly remarkable is that this model was developed at a cost of just $5.6 million, significantly lower than comparable models from competitors. For example, it cost $100 million to develop ChatGPT 4.0. 

More importantly, DeepSeek's commitment to open-source development and its cost-effective innovation has positioned it as a formidable challenger in the AI industry. It has also led to criticism by the US and a recent ban by South Korea as countries are questioning privacy. Despite this, the app continues to be the most downloaded app on the App store

Personally, I think it’s the Kodak moment of LLMs. I often teach my MBA students about the demise of Kodak and tell them to “disrupt or be disrupted”. Deepseek is clearly telling the other players that they will be disrupted.

The Rise of AI in Healthcare: A New Era of Possibility

As a practising surgeon, and fervent advocate for the integration of AI in healthcare, I have had the privilege of witnessing the profound transformation that AI is bringing to our field. Every day, clinicians, researchers, and administrators are faced with the Herculean task of managing an abundance of information—from patient records and medical literature to the intricate data from imaging and genomic sequences. 

The traditional methods of handling this data are simply no longer adequate. 

Enter AI, and more specifically, the advent of large language models (LLMs). These are not merely chatbots; they are powerful systems like ChatGPT, DeepSeek, and Qwen that possess the capability to revolutionise a new era of possibilities in healthcare.

Streamlining Administrative Tasks: AI is revolutionising this aspect of healthcare by automating tedious paperwork, thus liberating clinicians like myself to focus more on patient care. The relief of having AI manage these tasks is akin to having an invisible assistant who handles the minutiae, allowing me to channel my energy and expertise where it truly matters.

Democratising Knowledge: One of the most heartening aspects of AI in healthcare is its potential to democratise knowledge. In underserved communities and remote areas where access to medical expertise is limited, AI is bridging the gap. Large language models are making cutting-edge medical information accessible to those who need it most.

Enhancing Diagnostics: With AI, we can now glean real-time insights from an ever-expanding repository of medical literature and patient data. I have seen firsthand how AI can sift through this vast information at incredible speeds, providing us with critical insights that aid in quicker and more accurate diagnoses.

Personalising Treatment: No two patients are alike, and the era of one-size-fits-all treatment is rapidly becoming obsolete. AI allows us to delve deep into individual patient profiles, tailoring therapies that are uniquely suited to each person's specific needs.

As we journey further into this new era, each of these large language models brings its unique strengths to the table. DeepSeek's unparalleled data analysis, ChatGPT's conversational abilities, and Qwen's specialised knowledge each contribute to reshaping the future of medicine in their way.

DeepSeek: The Precision Partner for Medical Research

DeepSeek is a next-generation AI model designed to excel in precision and domain-specific applications. Unlike general-purpose LLMs, DeepSeek is tailored for industries like healthcare, where accuracy and reliability are non-negotiable. This groundbreaking technology leverages artificial intelligence to analyse vast amounts of medical data, providing unprecedented insights into patient care, diagnosis, and treatment. At its core, DeepSeek utilises advanced machine learning algorithms to sift through complex datasets, identifying patterns and correlations that human practitioners may overlook. This enables healthcare providers to make more accurate and timely decisions, ultimately improving patient outcomes. For instance, DeepSeek can analyse a patient's genetic information, medical history, and lifestyle factors to predict the likelihood of developing certain conditions, allowing for early intervention and personalised treatment plans.

Key Features in Healthcare:

1. Medical Literature Analysis: DeepSeek claims to have the capability to sift through millions of research papers, clinical trials, and case studies in seconds, extracting relevant insights for clinicians and researchers. For example, it can identify emerging trends in cancer treatment or highlight potential drug interactions. This accelerates the therapy and drug discovery process, bringing life-saving medicines to market more quickly and cost-effectively.
   

2. Clinical Decision Support: One of the most significant advantages of DeepSeek is its ability to enhance diagnostic accuracy. By integrating with electronic health records (EHRs), DeepSeek provides real-time recommendations to clinicians. Moreover, DeepSeek can identify subtle anomalies that may indicate the presence of diseases like cancer or cardiovascular conditions. This not only aids in early detection but also reduces the risk of misdiagnosis, which can have life-threatening consequences.
   

3. Patient Communication: In addition to its clinical applications, DeepSeek empowers patients to take a more proactive role in their healthcare journey. DeepSeek’s natural language processing (NLP) capabilities enable it to generate clear, empathetic responses to patient queries. This is particularly valuable in telemedicine, where patients often seek quick, accurate answers.

Potential Impact

By lowering the barriers to AI adoption, DeepSeek could enable a broader range of healthcare providers to implement AI-driven diagnostics, predictive analytics, and personalised treatment plans, ultimately improving patient outcomes across diverse settings.

ChatGPT: The Pioneer of AI in Healthcare

ChatGPT, developed by OpenAI, is nothing short of a revolutionary tool in healthcare, showing extraordinary versatility. Since its inception, it has become an invaluable asset and a pioneer in AI in various sectors, healthcare being one of its most impactful areas.

Key Features in Healthcare:

Administrative Efficacy: As someone engaged in the healthcare field, I find immense value in ChatGPT's ability to streamline administrative tasks. Automating mundane tasks like scheduling appointments, creating clinical notes, and coding insurance claims significantly reduces administrative burdens and minimises errors.

Medical Education: ChatGPT is a fantastic resource in medical education, as it simulates patient interactions, simplifies complex concepts, and even generates quiz questions for self-assessment, transforming the learning experience for medical students and professionals alike. This tool is indispensable for anyone aspiring to deepen their understanding of medical sciences.

Mental Health Support: While not a substitute for human therapists, it provides immediate, empathetic responses to individuals in distress. It’s particularly useful in regions with limited access to mental health care, offering a preliminary layer of support.

Potential Impact:

By integrating ChatGPT into healthcare systems, providers can improve patient communication, streamline operations, and support clinicians with up-to-date medical information, contributing to more efficient and effective care delivery.

Alibaba’s Qwen: Revolutionizing Healthcare with Multimodal AI

Alibaba’s Qwen is a multimodal AI model that goes beyond text, integrating images, videos, and even genomic data. This makes it uniquely suited for healthcare, where multiple data types are often needed to make informed decisions.

Key Features in Healthcare:

1. Enhanced Diagnostics and Treatment: Qwen2.5 Max's ability to analyse vast amounts of medical data, including patient records, research papers, and clinical trial results, can lead to more accurate and efficient diagnoses. It can also assist in developing personalised treatment plans tailored to individual patient needs.

2. Medical Research: Qwen2.5 Max can assist in medical research by analysing large datasets to identify patterns, trends, and insights that can lead to breakthroughs in our understanding of diseases and their treatment.

3. Telemedicine and Remote Monitoring: Qwen’s ability to process video and audio data makes it ideal for telemedicine. It can analyse patient videos for symptoms, monitor vital signs remotely, and even detect early signs of conditions like Parkinson’s disease through speech patterns.

Potential Impact:

Qwen AI has the potential to revolutionise healthcare by integrating multimodal data, including images, videos, and genomic information, to enhance diagnostic accuracy and personalise treatment plans. Its applications range from medical imaging analysis to genomic medicine and telemedicine, significantly improving patient outcomes and accessibility to advanced care, especially in underserved areas. Comparing the Models: Strengths and Limitations

The Road Ahead:

As someone deeply invested in AI’s role in healthcare, I see ChatGPT, DeepSeek, and Qwen as complementary tools with distinct strengths. ChatGPT's conversational agility makes it ideal for patient education and mental health support. DeepSeek, with its roots in China, excels in medical imaging analysis. At the same time, Qwen, integrated with Alibaba's cloud infrastructure, stands out in drug discovery and genomic data processing, already accelerating research timelines in early trials.

What excites me most is the collaboration these models could foster. Imagine ChatGPT triaging patient queries, DeepSeek analysing scans, and Qwen optimising treatment plans—all feeding insights into a unified system. But challenges loom. Bias in training data, like the underrepresentation of non-European genomes, risks worsening health disparities. Regulatory hurdles also persist: while Qwen aligns with China's PIPL laws, global standards for AI diagnostics remain fragmented. Data privacy concerns are also a major hurdle in the global adoption of these models. 

By 2030, the AI healthcare market is projected to hit $188 billion, but success hinges on ethical guardrails. We need hybrid models where AI handles data crunching while clinicians make final decisions. To conclude, these models aren't replacements—they're partners. Their future lies in humility: acknowledging limitations, prioritising transparency, and centring patient voices. As we innovate, let’s ensure AI doesn’t mirror healthcare’s existing inequities but helps dismantle them. 

Stay tuned for next week's edition of AI Horizons, where we will explore the role of AI in breast cancer.

Subscribe | Engage | Transform

Join 50,000+ pioneers at the nexus of AI and healthcare.