QuantDiagnose
Combing "Quant" and "Diagnose"
The problem QuantDiagnose solves
The hybrid autism detection model solves the challenge of identifying autism spectrum disorder (ASD) in individuals more accurately and efficiently. Early and reliable detection of autism can greatly enhance the outcomes for individuals by enabling timely intervention and support.
This model can be used by healthcare professionals, researchers, and institutions to assist in early diagnosis of autism, which is crucial for planning effective treatment and support. The hybrid model integrates classical and quantum-inspired methods to improve prediction accuracy, making the process of autism detection faster, more reliable, and scalable compared to traditional methods. It enhances diagnostic processes by combining the strengths of classical Convolutional Neural Networks (CNN) and more advanced machine learning techniques. This hybrid approach results in higher accuracy and reduced false positives, streamlining diagnosis and allowing for more informed medical decisions.
Challenges we ran into
One significant challenge was integrating the hybrid quantum-classical model for autism detection. Initially, we encountered issues with converting quantum circuit outputs to TensorFlow-compatible types, leading to errors in training. Additionally, setting up the correct architecture for the hybrid model, combining CNNs with quantum circuits, was complex, as it required careful tuning of the number of qubits and model layers to ensure compatibility and performance.
We overcame these hurdles by simplifying the quantum circuit using fewer qubits and utilizing StronglyEntanglingLayers in PennyLane to stabilize the model. Debugging TensorFlow errors and fine-tuning hyperparameters like learning rates and batch sizes also helped to improve the model's overall performance.
Tracks Applied (1)
Quantum Machine Learning
Technologies used