How Our Student Saksham Built an AI Antibiotic Resistance System in High School

Introduction: High School Student built Antibiotic Resistance System

Antibiotic resistance is called the “silent pandemic”. According to the CDC, it contributes to 1.27 million deaths globally every year AI-Driven Antibiotic Systems. It’s one of the greatest medical challenges of our time.

You’d think solving this problem requires a PhD and a million-dollar lab. But here’s the surprise: Saksham, a high school student at BetterMindLabs.org, built an AI-driven prototype to explore how machine learning could fight antibiotic resistance.

His work didn’t create a clinical solution—that takes decades—but it did showcase:

• Intellectual curiosity.

• Technical skills in AI and biology.

• Leadership and persistence.

• A project portfolio that stands out for colleges and scholarships.
  

This blog shares Saksham’s journey : what he built, how he built it, and how your student could follow a similar roadmap.

Why Antibiotic Resistance Matters

• Public Health Risk: WHO lists AMR (antimicrobial resistance) as one of the top 10 global health threats.

• Cost: Resistant infections cost billions annually in longer hospital stays and treatments.

• Innovation Gap: Only a handful of new antibiotics have been approved in the last 30 years, AI-Driven Antibiotic Systems.
  

AI is being explored to:

• Discover new drugs faster (MIT used AI to design “Halicin,” a new antibiotic against resistant bacteria).

• Predict resistance patterns from microbial genomes.

• Support clinicians in prescribing the right drugs at the right time.
  

Saksham’s AI Antibiotic Resistance Project

Saksham’s system combined four modules, inspired by cutting-edge research but scoped for a student-friendly prototype:

1. AI in Antibiotic Discovery

• Tools Used: Graph Neural Networks, Generative AI (diffusion models).

• Goal: Generate new molecule candidates with antibacterial properties.

• Research Parallel: MIT’s AI pipeline discovered novel antibiotics active against MRSA and E. coliAI-Driven Antibiotic Systems.

• Student Angle: Saksham used open chemical datasets (ChEMBL, ZINC) and simulated “molecule suggestions.”

(BetterMind Labs student project parallel: “ChiralAI,” an AI system built by Alexei Manuel for healthcare diagnostics.)

2. Compound Optimization

• Tools Used: Multi-objective optimization scoring.

• Goal: Filter molecules for safety, solubility, and synthesizability.

• Why It Matters: 90%+ of drug candidates fail clinical trials because they’re unsafe or impracticalAI-Driven Antibiotic Systems.

• Student Angle: Saksham learned to design scoring functions balancing potency vs. toxicity.
  

3. Predicting Antimicrobial Resistance (AMR)

• Tools Used: Random Forest, XGBoost, Neural Networks.

• Dataset: Publicly available pathogen genome + resistance databases.

• Result: A model with AUC ~0.9 in predicting whether a bacterial strain was resistant to a given antibioticAI-Driven Antibiotic Systems.

• Healthcare Impact: In real life, this could speed up treatment selection.

• Student Angle: Focused on learning about imbalanced datasets and the need for multiple evaluation metrics (ROC-AUC, Precision, Recall).
  

4. Clinical Decision Support Prototype

• Tools Used: Streamlit, Python APIs.

• Goal: A demo app where a doctor could input patient + pathogen data, and the AI suggests antibiotics with probabilities of success.

• Ethical Guardrails: Every screen said “For educational purposes only — not for clinical use.”

• Research Parallel: Clinical decision support systems like KINBIOTICS are being trialed to help ICU doctors pick antibiotics fasterAI-Driven Antibiotic Systems.
  

Week	Focus Area	Activities	Outputs
1	Literature Review	Read CDC + WHO reports on AMR; summarize 3 key threats	1-page summary
2	Data Collection	Explore open datasets (ChEMBL, ZINC, pathogen genomes)	Dataset notes
3	Drug Discovery AI	Train a generative model to create molecule structures	Notebook with sample molecules
4	Optimization	Apply scoring rules for potency, toxicity, solubility	Ranked molecule list
5–6	AMR Prediction	Train classifiers on pathogen-resistance data	Model metrics (ROC-AUC, F1)
7	Clinical Support App	Build Streamlit demo with input/output	Prototype app
8	Ethics & Documentation	Add disclaimers, discuss limitations, cite research	Final report + README
9–10	Portfolio Polish	Record demo video, upload to GitHub, write reflection	Portfolio-ready project

What Parents Should Know

• Safety First: These projects must use public datasets only. No hospital data, no patient information.

• Educational Scope: Always frame as learning prototypes, not medical tools.

• Mentorship Helps: Programs like BetterMindLabs.org give guardrails, weekly check-ins, and ensure projects stay rigorous yet safe.

Other Student Projects at BetterMind Labs

Saksham isn’t alone. Other high schoolers built projects like:

• Stroke Detection AI (Aryaman) – Using imaging and tabular health data for early warning.

• ChiralAI (Alexei) – AI in healthcare diagnostics (see video).

• Fraud Detect AI – Predicting fraudulent transactions.

• Finance Buddy AI – Personal finance tracking and budgeting.

• Warehouse Buddy – AI optimization for inventory and logistics.
  

These projects show the range of what teens can achieve when curiosity meets mentorship.

Conclusion

Saksham’s AI Antibiotic Resistance System wasn’t about solving the global health crisis in one step. It was about showing what’s possible when a high school student takes initiative, works with mentors, and uses AI to address real-world problems.

For families, the takeaway is this:

• Yes, your teen can do this.

• Yes, it builds credibility for college, scholarships, and careers.

• Yes, mentorship matters. Programs like BetterMindLabs.org provide the safe structure and expert guidance to turn big ideas into portfolio-ready projects.
  

Because innovation doesn’t just come from labs—it can come from motivated high school students like Saksham.