LLM-Driven Allergy Symptom Tracker for Personalized Health Management

Allergy Symptom Tracker: How a Large Model Helped Me Win the "Sneeze War"

FirstLet Me Tell You About My Allergy Hell
Every springI feel like I've lost a battle—my nose is as clogged as if it were filled with cementmy sneezes are so violent that even I'm scaredand my eyes itch so badly I want to rip them out and wash themThe most frustrating part is that I have no idea whether it's pollendust mitesor the mango I ate yesterday that's causing itI've tried various allergy medicationsbut sometimes they work and sometimes they don'tmaking me feel like I'm just guessingThen a friend recommended trying this allergy symptom trackerclaiming it's powered by a large model and can help me find patternsAt firstI thought"Can a piece of software be smarter than a doctor?" But after using it for a weekI was completely sold!

What Exactly Can This Thing Do?

Simply putit's an allergy symptom diary plus an intelligent analystEvery morningafter lunchand before bedyou open it and record the discomfort you've had that day—for example"sneezed 5 timesnose a bit itchyeyes red." Then you check off suspected triggers (high pollen concentrationate peanut butterjust had a major house cleaning)and what medication you took (one loratadine tablet)The large model will combine all these bits of data and tell you"Dudeyour symptoms worsened for three consecutive days last weekall related to peak pollen levelsI suggest you stay indoors for the next few days." Pretty thoughtfulhuh?

How to Install? Super SimpleThree Steps

Since it's an open-source projectyou can get it directly from GitHubFirstmake sure your computer has a Node.js environment (if notdownload it from the official site—it only takes a few minutes)Then open your terminal and type this command:

npx skills add https://github.com/itsoso/health-llm-driven --skill allergy-symptom-tracker

It will automatically download and configureIf you like to tinkeryou can also clone the entire repository:

git clone https://github.com/itsoso/health-llm-driven.git
cd health-llm-driven/backend/skills/allergy-symptom-tracker
npm install

Then start the service with npm start and open http://localhost:3000 in your browserOf courseyou can also run it directly on theplatform with one click—effortless and worry-free.

First Thing to Do: Create an Allergy Profile

When you first open itit will ask you to fill in personal informationsuch as ageknown allergensand whether you have a history of asthmaDon't be lazy—the more detailed you fill it inthe more accurate the large model's analysis will beThen you can start recording! I usually record three times a day: morning after waking upafternoon after picking up the kids from schooland evening before bedOn weekendsI also record anything special I ate.

After about five days of recordingI discovered a pattern: every time I ate hot pot (especially with lots of mushrooms and bean sprouts)I would sneeze the next dayBeforeI thought it was just the spiciness of the hot pot that irritated mebut the tracker told me that mushrooms and bean sprouts may contain histamineand chili itself is a histamine releaser—a double whammyFrom then onI only order clear broth for hot potand my symptoms have been cut in half.

Features That Wowed Me

• Voice Input: Too lazy to type? Just say"Today I have a runny nose and coughworse after breakfast," and it automatically converts to textA lifesaver for the lazy.
• Trigger Suggestions: Based on your past recordsit intelligently recommends possible new triggersFor exampleI initially didn't pay attention to pollenbut it analyzed that my symptom peaks coincided with the local pollen calendar and proactively suggested adding a "pollen" tag.
• Alert Reminders: You can set a thresholdFor instanceif your symptom score exceeds 6 for two consecutive daysit pops up a notification: "Warning! You may be exposed to high doses of allergensIt is recommended to take allergy medication and consult a doctor immediately." One time I almost forgot to take my medicationthanks to its reminder.
• Visual Reports: Every week it generates a radar chart showing the distribution of symptom types (sneezingnasal congestionskin rashetc.) and severity changesYou can also export it as a PDF to show your doctor—much more reliable than handwritten records.

Is the Large Model Analysis Just Hocus-Pocus?

To be honestI was skeptical at first about how accurate the AI could beBut after using itI found its analysis actually makes senseFor exampleI recorded that "my lips go numb every time I eat a banana." It didn't simply conclude a banana allergy; insteadit combined the "numb lips" symptom with "recent cold" data I also recordedand determined it might be angular cheilitis caused by the coldunrelated to bananasLater I went to the doctorwho also said it wasn't a food allergySee? The AI saved me a trip to the clinic and a blood test feeBut if you have severe allergiesdon't skip the doctor—this tool is just an aid.

A Few Pitfalls and Notes

• Regular recording is key: If you record sporadicallythe model can't find patternsI set an alarm to record at fixed times every day.
• The large model needs an internet connection: Because it calls a large language model APIyou must stay onlineIf offlineonly local recording works—no smart analysis.
• Privacy concerns: Your symptom data is sent to the model serverIf you care a lot about privacyconsider setting up your own private large model (e.grun a local LLM with Ollama) and modify the configurationThis project supports custom API URLswhich is very user-friendly.
• Don't blindly trust AI advice: The AI analysis is for reference onlyAlways follow your doctor's guidance for medication and diagnosisI use this tool only for auxiliary management; I never dare to replace formal medical care.

More Than One Use Case

Besides pollen allergiesit is also suitable for:

• Food allergies (milkeggsseafoodetc.)
• Seasonal rhinitis
• Contact dermatitis (cosmetic allergiesmetal allergies)
• Even pet hair allergies

My dog sheds a lotand I've always suspected I might be allergicI used the tracker for a week and found that I start sneezing 15 minutes after holding the dogand it positively correlates with the amount of hair my dog sheds