Trial Success Metric System using Quantum Computers
Team QUBO
Archana Iyer
Soham Chatterjee
Why Quantum Computers?
- Computation: Since quantum bits or qubits can exist in both states 0 and 1 as compared to a classical computer which can only exist either 0 or 1, the computation capabilities for a quantum computer is exponentially increased.
- Quantum Search: The potential of quantum computers has taken a leap to understand problems in a given search space.
- Solving Optimisation problems
- Quantum Cryptography
Trial Success Metric System
- Key Feature of the TPO. It helps in predicting the likelihood of a Trial being successful will be hugely beneficial during Trial Design.
- New MIT Study Puts Clinical Research Success Rate at 14%. Only 14% of all drugs in clinical trials eventually win approval from the FDA
- Several factors which affect this include :
- Recruiting patients: Finding the right patient for a clinical trial. An increase in attrition rates of patients during a trial is also high.
- Lack of Infrastructure or Sites
- This further leads to lower chances of the drug reaching the market.
Data Used
Data used: 3k Citeline trial data that focuses on Oncology Leukemia. Input Parameters considered were Trial_phase (categorical), sponsor_academic (1 or 0), sponsor_cooperative_group (1 or 0), sponsor_government (1 or 0), sponsor_industry (1 or 0), sponsor_not_for_profit_funding_entity (1 or 0), sponsor_top_20_pharma (1 or 0), sponsor_all_other_pharma (1 or 0 )
Output : Whether the trial would be successful or not-i.e. can be either 0 or 1 .
The architecture of our System
RBM running on a Quantum Computer
Logistic Regression Model
Trial Success/Failure
Demo
Advantages of this Method
- Tapping one of the most technologically advanced computing technique: Quantum Computing
- QC took less than 30 seconds to train for 4 Epochs as compared to Classical Computer which takes more time
- First application of Quantum for Clinical Trials (as per our research)
- Train on Classical and Quantum Computer simultaneously
- Further work could be done in understanding clinical drugs, drug-drug interaction and drug discovery
Bringing DaLIA to the Edge
Why Bring DaLIA to the Edge?
- Immense amount of data produced by IoT devices running parallel with the sheer increase in IoT devices across the world
- Real-World Interactions to DaLIA: DaLIA can be a tool that can be expanded and be integrated for conversations with people and their devices
- DaLIA to connect the less privileged
DOSA: Dalia on Small Area
- A device that helps to run machine learning models on the edge with DaLIA that acts as an interface to the public
-
Currently, this device supports two features
- Skin Cancer Detection through Image Processing
- Detecting sickness sounds in patients
- Use case: This device could be employed with doctors in rural areas where DaLIA could help act as an interface to patients and the deep learning models can help identify in real time whether the patient is suffering from skin cancer/sickness
Architecture of the System
UP board Microcontroller
Sickness and Cancer detection Neural Network models
Demo
Q & A
deck
By archana iyer
