Safe Scan: Decentralized Intelligence for Early Cancer Detection

Cancer disrupts life, showing up in the most unexpected places at the most unexpected times. While it can be immediately devastating, the point that lies between early detection and late detection is often crucial and determines whether or not it is a death sentence. 

However, the equipment that can be used to detect it early is often not made available to those who most require it. Already, AI research demonstrates some level of accuracy in detecting certain types of cancer, and these findings are rarely published outside of the academic community.

Hence, Safe Scan (Subnet 76) embarked on eliminating that obstacle by making the process of cancer detection accessible, affordable, and decentralized, i.e., not run by one institution but by thousands of contributors.

The Rise of Safe Scan

The interest was instant when Safe Scan entered the medical world. This was the team that appeared at MedTech Meetup Poland in Katowice and later at Science Non Fiction with the National Oncology Institute of Maria Sklodowska-Curie. Those events proved that a decentralized approach to medical work was not only technologically interesting but also required.

The project subsequently matched its multi-skin-disease recognition competition with the MILK10k Benchmark, which put the Bittensor-trained models on an equal footing as global research benchmarks. Comparison with state-of-the-art systems is anticipated in the near future for melanoma.

Breakthrough Findings

The C2S-Scale 27B model marked one of the largest achievements of the project, as it identified a possible method for reawakening dormant tumors and causing them to respond to immunotherapy.

The Tricorder-3 Benchmark

In order to substantiate these findings, Safe Scan created the Tricorder Challenge, which happens to be in its third round. 

Documentation:

• https://github.com/safe-scan-ai/cancer-ai/blob/main/DOCS/competitions/TRICORDER-3.md
• https://github.com/safe-scan-ai/cancer-ai/blob/main/DOCS/competitions/TRICORDER-2-TO-3-DISEASE-MAPPING-TRANSITION.md
• https://github.com/safe-scan-ai/cancer-ai/blob/main/DOCS/competitions/TRICORDER-3-DISEASE-MAPPING.md

When it is treated promptly, the survival of Melanoma is more than 99 percent; whereas, when it is treated later, the survival is less than 30 percent. Basal Cell Carcinoma and Squamous Cell Carcinoma are similar to each other, and this is why it is important to detect them at the earliest stage. Owing to this, the assessment assigns a triple weight to malignant lesions and a single time weight to benign cases. The performance in real-time can be seen on the public dashboard.

How It Works

• The AI is an open-source cancer detection algorithm that has been trained on large medical datasets. 
• Anyone can also be a miner and receive TAO as they add processing power that reinforces cancer detection around the globe.
• Safe Scan also comes with a scientist-rewarding model. TAO rewards developers, researchers, and machine learning experts with the best and most useful algorithms. The higher the performance of the model, the more it earns. This guarantees that the algorithms that are applied in real life are the strongest.
• Those algorithms do not remain theoretical. As soon as an algorithm has been effective during the competitions, it is implemented into software that is utilized in real diagnostics of a cancer.
• The more nodes are added, the faster the diagnosis will be completed, and the network can become capable of supporting even more types of cancer, like breast, lung, and, sometimes, brain cancer.
• Privacy is at the core. Data relating to patients remains anonymous, and its processing is decentralized, so a specific organization cannot store or manipulate sensitive data.
• The other key focus of Safe Scan is strategic collaborations with hospitals, MedTech companies, and universities. Safe Scan is shifting to FDA compliance in the US and MDR/CE compliance in the EU.

Clinical Trials and Deployment

The project is in the process of conducting authorized clinical trials of its skin cancer detecting application, SELFSCAN. The product will be taken to the Class II medical device registration stage in the United States and Europe after trials. Future breast and lung cancer detection tools are to be conducted in the same direction.

Growth Strategy and Market Impact.

Safe Scan is projected to accomplish its goal of engaging more than one million users of this app in 18 months by collaborating with cancer foundations, affiliate programs, and exposure in the app stores. 

The top ten miners are rewarded through competitions, with 50 percent being awarded to the first place and decreasing to 1 percent for the ninth and tenth places. 

The Team Behind the Work

The team blends machine learning, mobile development, software design, and business leadership. They are close friends as well as collaborators, united by the belief that decentralized AI should save lives, not sit in academic files.

• Github repo (showing more details about the team): https://github.com/safe-scan-ai/cancer-ai

Why SN76 Matters

Since purchasing SN76 tokens is contingent on continued subnet activity, holders of the alpha support the network, driving Safe Scan’s greater effort to enhance early cancer detection.

Conclusion

Safe Scan is at that point where research meets community power and medical desire. It aims at making cancer scans more accessible to those who need them while creating a system that pays those who can enhance models related to it. 

Since more attention is focused on it while trials are ongoing and with a bright future ahead of the project, Safe Scan is indeed at a point where its actual work relates to a community that would like to see early cancer detection, just like any other ordinary medical practice.