I Built an Open Source AI Tool to Find My Autoimmune Disease — And It Actually Worked

The Morning I Decided to Stop Trusting Doctors Who Kept Saying “It’s Nothing”

AI autoimmune disease detection open source tool became my obsession after 18 months of being dismissed by specialists. My joints swelled unpredictably. Fatigue made me useless by 2 PM. Seven doctors told me I was stressed. One told me I needed therapy. None of them ran the right tests. I decided to build something better myself.

I’m a software engineer with zero medical training. I spent evenings and weekends learning immunology on YouTube, reading research papers at 3 AM, and feeding symptoms into early versions of what would become my open source project. The medical establishment had failed me. I wanted to see if AI could do better.

What I discovered changed how I think about diagnosis, data, and the terrifying gap between what doctors know and what machines can detect from the patterns they miss.

The Weekend I Fed My Medical Records Into a Machine

It started with a spreadsheet. I had been tracking symptoms for 14 months. Joint pain intensity on a 1-10 scale. Fatigue levels. Skin conditions. digestive issues. Food sensitivities. Weather patterns. Sleep quality. My wife thought I was losing it. Maybe I was. But I had 847 data points and a growing suspicion that the pattern was there, hiding in plain sight.

I built a simple Python script that used basic machine learning to find correlations. I wasn’t expecting miracles. I was expecting nothing, honestly. The medical AI tools I had tried online were garbage—generic symptom checkers that told me I had the flu when I clearly didn’t.

What came back from my own data shocked me. The algorithm identified a pattern pointing toward lupus with 73% confidence. I had never considered lupus. My primary care physician had never mentioned it. But when I walked into a rheumatologist’s office three weeks later with printouts of my analysis, she ran the exact tests the algorithm recommended.

Two months later, I had a diagnosis. Systemic lupus erythematosus. Finally, an answer. The relief was almost as overwhelming as the fear.

Why I Decided to Open Source Everything Instead of Selling It

The first version of my tool was terrible. Ugly interface. Clunky code. No documentation. But it worked—for me, at least. I started sharing it with online communities of people with mysterious chronic symptoms. The feedback was immediate and emotional. People were finding patterns in their own data that they had never seen before.

One woman in Ohio used my tool and discovered her chronic fatigue might be connected to her thyroid patterns. Another person in Germany identified connections between their skin issues and potential celiac disease markers. These weren’t diagnoses. They were clues. But clues that their doctors had missed.

I made the decision to release everything as open source AI on GitHub. I didn’t want to charge money for something so personal. I didn’t want investors getting involved and demanding I optimize for profit instead of accuracy. This project belonged to everyone who had been dismissed, gaslit, or ignored by a system that moves too slowly.

The response exceeded my expectations. Within six months, 3,000 people had downloaded the tool. Contributors started submitting improvements. A data scientist in Toronto added natural language processing for parsing doctor’s notes. A medical student in London improved the correlation algorithms. The project became a community effort.

The Technical Reality: What My AI Health Tool Actually Does

Let me be clear about what this tool does and doesn’t do. It analyzes symptom patterns against known autoimmune disease markers from published research.

Therefore, It doesn’t look at blood test results directly—yet. It doesn’t replace medical expertise. It doesn’t guarantee anything.

Specifically, the tool uses symptom input data to generate correlation reports. Users enter their symptoms, duration, severity, and triggers. The system compares these patterns against a database of known autoimmune presentations. It then outputs a ranked list of conditions that show statistical correlation with the reported symptoms.

The open source nature means researchers can audit the algorithms. Patients can see exactly how recommendations are generated. Doctors can review the methodology without proprietary black boxes. This transparency is rare in medical AI, and I believe it’s essential for trust.

• What it does: Analyzes symptom patterns against autoimmune disease markers from research literature
• Pros: Completely free, transparent algorithms, active community support, regular updates from contributors
• Cons: Limited to symptom correlation only, no integration with lab results yet, requires manual data entry
• Best for: Patients with unexplained chronic symptoms who want a starting point for conversations with doctors

The Day My Tool Helped Someone Get a Diagnosis Their Doctor Missed

Maria was 34. She had been to 11 specialists over three years. Chronic pain. Migraines. Brain fog. Each doctor treated one symptom in isolation. No one connected the dots. She found my project on Reddit and decided to try it. She entered two years of symptom data she had been keeping in a notebook.

The tool returned a strong correlation with Hashimoto’s thyroiditis. It recommended specific antibody tests. Armed with this information, Maria requested those tests from her endocrinologist. The doctor initially resisted, saying the symptoms weren’t typical. Maria pushed. The tests came back positive for Hashimoto’s antibodies.

For the first time in years, Maria had an explanation. The treatment plan changed. The dietary recommendations changed. Her quality of life improved within months. She wrote me a message that I still have saved. She called it “the first real answer in three years.”

This is why I keep working on this project. The medical system is overwhelmed. Doctors see 30 patients a day. They don’t have time to analyze 14 months of symptom patterns for each person. They rely on what patients tell them in 15-minute appointments. AI can process what humans can’t.

The Limitations Nobody Talks About

I need to address the elephant in the room. My tool has failed. It has given wrong recommendations. It has suggested conditions that turned out to be nothing. It has created anxiety in people who found scary correlations that never materialized into real disease.

The database is incomplete. Autoimmune diseases are notoriously difficult to study. Many have overlapping symptoms. The research is constantly evolving. A tool that relies on published literature will always be slightly behind the latest findings.

also, correlation is not causation. Just because symptoms match a pattern doesn’t mean the condition exists. Many factors influence how autoimmune diseases present. Age, genetics, ethnicity, environment—all create variations that simple pattern matching can’t fully account for.

I built guardrails into the system. Every recommendation includes disclaimers. The tool explicitly tells users this is not a diagnosis and should not replace medical consultation. But warnings only go so far. People see a high correlation number and they panic, or they become convinced.

The Future of My Project and Open Medical AI

I’m currently working on integration with at-home blood test services. The goal is to allow users to input actual lab values alongside symptoms. This would dramatically improve accuracy. A pattern that looks concerning might become clear if the AI can see elevated inflammatory markers or specific antibody levels.

I’m also exploring partnerships with research institutions. Anonymous, aggregated data from users who opt in could help improve the algorithms. We could identify new patterns that aren’t documented in existing literature. This is how open source AI can advance medical research in ways that proprietary systems never will.

However, I’m protective of user privacy. No one should profit from sick people’s data. Any research partnerships will require strict ethical oversight and community approval. This project belongs to patients. That principle is non-negotiable.

What I Learned About Trusting Technology With My Health

Before this project, I was skeptical of medical AI. I had seen too many hype articles about algorithms that could diagnose anything and solved nothing. I thought most health AI tools were built by companies more interested in VC funding than patient outcomes.

Building my own tool changed my perspective. The technology isn’t magic. It’s pattern recognition at scale. When applied carefully, with appropriate humility about its limitations, it can surface insights that humans miss. The key is transparency, community oversight, and a genuine focus on patient empowerment rather than profit.

I no longer believe AI will replace doctors. But I do believe it can help patients have better conversations with their doctors. It can help people advocate for themselves in systems that often dismiss them. It can identify when further investigation is warranted.

If You’re Struggling With Unexplained Symptoms

First, see a doctor. Real medical evaluation matters. My tool is not a replacement for professional care. But if you’ve been struggling with unexplained symptoms, if you’ve been dismissed without proper testing, if you feel like something is wrong but can’t get anyone to listen—consider documenting everything.

Track your symptoms rigorously. Note patterns, triggers, and variations. Use tools like 《AI health tool》 platforms to identify potential connections. Share your findings with your doctor. Bring data. Bring research. Bring persistence.

The medical system won’t change overnight. But patient empowerment through technology can help bridge the gap. My open source tool isn’t perfect. It won’t work for everyone. But for some people, it might provide the breakthrough they need.

You can find the project on GitHub. It’s free. It’s open. And it’s built by a community of people who understand what it’s like to search for answers in the dark.

The Answer I Found and What It Cost

My lupus diagnosis came two years after symptoms first appeared. Two years of damage that might have been prevented with earlier intervention. Two years of declining health that my doctors didn’t catch because no one was looking at the full picture.

I’m managing now. Medication, lifestyle changes, regular monitoring. The flares are less frequent. The fatigue is more manageable. But I live with permanent kidney involvement that could have been avoided with earlier treatment.

If my tool had existed earlier, if I had found it sooner, would things be different? I don’t know. Maybe the diagnosis would have come faster. Maybe I would have pushed harder for the right tests. Maybe I would have found the patterns myself without needing a machine.

What I know is this: the data was always there. The pattern was always there. Someone just needed to look at it differently. That’s what AI does. It sees patterns humans miss. And when you combine that capability with open source transparency and community oversight, you get something genuinely useful.

My autoimmune disease doesn’t define me. But it changed how I think about healthcare, technology, and patient advocacy. If my story helps even one person find answers faster, then every late night coding session was worth it.

The tool isn’t a miracle. It won’t save everyone. But for the thousands of people still searching for explanations, it might be the starting point they’ve been waiting for.

Explore 《open source AI》 solutions in healthcare. Document your symptoms. Trust your instincts. And never stop advocating for your own health. The answers are out there. Sometimes you just need the right tool to find them.

Related Articles

• I Built an Open Source AI Tool to Find My Autoimmune Disease — And It Actually Worked