DeepSeek AI: What Does It Mean For Healthcare?

In late January 2025, an event occurred that was termed as “AI’s Sputnik moment”, as it appeared that the US was overtaken as the leader in artificial intelligence (AI) development. The event in question was the launch of DeepSeek, an AI chatbot from a Chinese company that functions similar to the likes of ChatGPT. 

The main difference was that this newcomer’s tool was as powerful as its contemporaries but it was developed with significantly fewer resources. This led to tech investors to question the lead that US companies have in this space, leading to the stock market to get dragged down.

While DeepSeek is in its early stages, it certainly has implications for healthcare and it presents some challenges for its adoption in this space. In this article, we provide an overview of the technology and what it means for healthcare.

What is DeepSeek and what is the hype around it?

Based in the Zhejiang province of China, DeepSeek was co-founded in 2023 by its current CEO, Liang Wenfeng. Between 2023 and 2024, the startup focused on AI research, developing several AI models that they made available as open-source material. Then in January 2025, they released their first free large language model chatbot, the eponymous DeepSeek, which they claim to be on par with the OpenAI-o1 model. 

It quickly rose in popularity to become the most-downloaded free app on the US iOS App Store. But this was not only due to its high-ranking performance, but also how it achieved that performance. As the US has restricted the sale of advanced microchips to China, citing security concerns, DeepSeek’s model had to be developed on relatively inferior technology and at a lower cost. 

Even with lesser resources, the company has achieved much of what other US tech companies have in this space. Furthermore, DeepSeek is cheaper to use; depending on the task, it can be 20 to 50 times cheaper to use than the OpenAI-o1 model. This brought into question the reasoning for US companies to invest billions in AI development, resulting in a hit in market shares of major players such as NVIDIA.

Furthermore, DeepSeek has open-sourced its DeepSeek R1 model that underpins its chatbot. With such accessibility and quality, DeepSeek has also piqued the interest of other researchers and startups. They can leverage a powerful AI tool without a significant financial burden. 

DeepSeek’s implications for healthcare

It’s been less than a month since the launch of DeepSeek’s chatbot, so its potentials have yet to be fully realised. In the strictly-regulated healthcare space, it is likely to take longer before we see any implementations. Despite being in its early stages, DeepSeek has implications in this field which are worth contemplating.

Being cost-efficient without compromising on quality, DeepSeek could make AI more accessible in the healthcare space. As its model is open-source, even smaller hospitals and research groups now have access to a sophisticated model at zero cost. Those with enough computational resources can even run the AI model offline on local systems for a secure and private solution. It is free to be used to tinker around and be repurposed to meet their needs. 

This can foster innovation for various healthcare use cases. For example, it can assist medical professionals in providing personalized treatment recommendations based on patients’ medical histories. As DeepSeek R1 is an efficient and low-cost model, it could also reduce healthcare delivery time and cost.

Concrete examples are not as widespread at this stage, but it is practically a certainty that the AI community is working on healthcare uses for DeepSeek. One early example comes from the WhyHow.AI team. They launched PatientSeek, an open-source AI tool based on DeepSeek’s model that has been fine-tuned on accessible datasets. It can create medical summaries and answer related questions to users. DeepSeek’s reasoning capabilities even enables the tool to discover correlations between patient records and external events.

Concerns and challenges with this new AI model

Generative AI tools like DeepSeek have undeniable potential to enhance healthcare delivery. However, concerns have been raised by the fact that the DeepSeek chat app stores data in Chinese servers. This has led US lawmakers to push for a ban of the app on government devices. Australia has already done so, citing its unacceptable risk to national security.

Politics aside, further concerns were drawn by Wiz analysts. They discovered security vulnerabilities which indicated potential information exposure. Such lack of stringent data integrity could make this particular app almost impossible to be used in healthcare.

Of note, these concerns only apply to DeepSeek’s own app. The AI model itself, being open-source, can be downloaded, repurposed and used for individual purposes while running on servers of choice. This approach offers an opportunity to address the crucial aspect of transparency for AI in healthcare, which other models like OpenAI do not offer. Furthermore, this open-source nature enables independent developers to refine, scale and improve the model as needed over time. 

As with all technology adoption in healthcare, DeepSeek-based tools need to be rigorously evaluated to ensure long-term security, reliability and privacy. Regulators have to step up to the task as we are likely to come across such tools, whether it is from DeepSeek or other AI models.

Written by Dr. Bertalan Meskó & Dr. Pranavsingh Dhunnoo