Web Application for Alzheimer’s disease modeling and drug discovery


Dementia is a broad umbrella term for a vast majority of neuro cognitive disorders including Parkinson's disease, Huntington's disease, and Alzheimer’s disease. It has been estimated that using the Reisberg Scale [1], the average expected life time for someone depends on the stage of dementia they are at. Based on the results from [1], expected life expectancy is more than 10 years for very mildly demented and 10 years for mildly demented patients. The scale goes down vastly from moderately demented (3 to 8 years), moderately severe demented (1.5 to 6.5 years), severe demented (less than 4 years) and very severely demented (less than 2.5 years). So, an early detection of dementia will be very beneficial for slowing down the progression of the disease to severe stages where any treatment becomes impossible. Problem - 1: We will be primarily focussing on early detection of the mildly demented stage, so that effective treatment can be started to slow down the progression.

Alzhiemer’s disease is a type of dementia that affects memory, thinking and behavior. It is also the most common cause of dementia [2]. Alzhiemer’s disease (AD) accounts for 60-80% of dementia cases. AD is also a progressive disease meaning that in early stages the memory loss is mild but with late-stage AD the individuals lose the ability to carry on a conversation and respond to their environment.

Alzheimer's has no cure, but some treatments demonstrate that removing beta-amyloid plaques, one of the important reasons of Alzheimer’s disease due to their accumulation between neuron connections, from the brain reduces cognitive and functional decline in people living with early Alzheimer’s. Other treatments can temporarily slow the worsening of dementia symptoms and improve quality of life for those with Alzheimer's and their caregivers.

Based on the reference paper [3], the progression of Alzheimer’s disease can be broken down into 3 general stages - Cognitive normal, Mild Cognitive impairment (MCI) and dementia. It is also more accurately described in a 7 stage model as mentioned in [3].

  1. No impairment

  2. Very mild cognitive impairment

  3. Mild cognitive impairment (MCI)

  4. Moderate cognitive decline (early stage dementia)

  5. Moderately severe cognitive decline (early mid-stage dementia)

  6. Severe cognitive decline (late mid-stage dementia)

  7. Very severe cognitive decline (late -stage dementia)

It is also mentioned in [3] that Alzheimer’s disease diagnosis is made at stage 2 or 3 of the above model. In this stage the individual can still function independently and free of dementia. This phase of Alzheimer’s disease can be detected well before the onset of dementia symptoms, up to 8 years in some cases. Problem - 2: Thus we need an approach to Early detect the Mild Cognitive impairment (MCI) stage so that it will be possible to adopt treatments to slow down the progression of this disease.

Alzheimer’s disease is caused by the accumulation of plaques in between the neural connections of the brain. These plaques prevent the transmission of neural signals between different neurons thereby removing their connection causing loss of memory.

Through various years of research, a hypothesis has been validated that Acetylcholinesterase (AChe) [4] inhibitors are effective in slowing down (or reversing) Alzheimer’s disease and Myasthenia gravis. Alzheimer’s Disease is characterized by lower levels of Acetylcholine than normal cognition. Acetylcholine is a neurotransmitter that breaks down into acetate and choline.It’s role is to terminate neuronal transmission and signaling between synapses. Acetylcholinesterase (AChE) inhibitors result in higher concentrations of acetylcholine and better communication between neurons.This can temporarily improve or stabilize Alzheimer’s disease symptoms.Current AChE inhibitor treatments available for AD include Donepezil, galantamine and Rivastigmine.

Problem - 3: The main problem here is that, each time a newer molecule is constructed, the properties need to be tested empirically causing a huge delay in drug testing. To solve this issue, it would be better to have a computational model which can identify whether the drug can target the disease or not. Problem - 4 :It would also be better to have generative models to create newer drug molecular compounds.

This is where our project comes to play. Based on the above problems, we propose 4 different hypotheses to solve each of them efficiently.

Hypothesis Proposed

Hypothesis - 1.1 : Early detection of dementia

We use the structural MRI scan images of the brain provided by Kaggle's [5] open source dataset and train a computation model to predict whether the patient is demented or not. Then, the model categorizes the MRI into a very mildly demented, mildly demented, or moderately demented stage based on the severity. This solves the first problemwhere an early detection of mildly demented stage is beneficial for treatment.

Hypothesis - 1.2 : Early detection of Mild Cognitive Impairment (MCI)

The dataset used to train our model is taken from ADNI [6] (Alzheimer’s disease neuroimaging initiative). The ADNI dataset is an open sourced dataset containing various reports, scan images and other features for many cognition based diseases. We will be using this ADNI dataset and gather the structural MRI scan images of various patients which are being grouped together into 3 classes - Cognitive Normal, Mild Cognitive Impairment and Alzheimer’s disease. Around 3370 images containing the Raw structural MRI scans, have been gathered from this ADNI dataset for training the model. This solves the second problem where an early detection of MCI stage will be beneficial to slow down the progression of Alzheimer's disease which can later become dementia.

Hypothesis - 2.1 : Drug Target Prediction for Alzheimer’s disease

The dataset is curated from the vast section of open source drug target dataset available in ChEMBL [7] database . ChEMBL is a manually curated database of bioactive molecules with drug-like properties. It brings together chemical, bioactivity and genomic data to aid the translation of genomic information into effective new drugs.

The database is queried for single protein targets based on their bioactivity towards inhibiting this AChe inhibitor. The bioactivity data for Human Acetylcholinesterase (AChe inhibitors) which has target values has been selected for model training. The main parameter used here for training the model is the IC50 value.

The bioactivity data is the IC50 value. Compounds having values of less than 1000 nM will be considered to be active while those greater than 10,000 nM will be considered to be inactive. As for those values between 1,000 and 10,000 nM will be referred to as intermediate. The dataset is prepared with the molecular ID, the canonical smiles representation of the molecule and their corresponding IC50 value.

Using this dataset, a suitable model is trained which can efficiently predict the bioactivity of a newer unknown molecule towards the AChe inhibition. This solves the third problem where the time required to empirically determine the compounds bioactivity is reduced using a computational model with higher accuracy of estimation.

Hypothesis - 2.2 : Drug molecule generation for Alzheimer’s disease

We use the same dataset from ChEMBL [7] to query for single protein targets based on their bioactivity towards inhibiting this AChe inhibitor. The bioactivity data for Human Acetylcholinesterase (AChe inhibitors) which has target values has been selected for model training. The main parameter used here for training the model is the IC50 value.The bioactivity data is the IC50 value. Compounds having values of less than 1000 nM will be considered to be active and taken for model training.

We will train a generative model which takes in these molecules and generates newer molecules towards AChe inhibition.


Classical Deep learning models

It is a method in artificial intelligence (AI) which teaches computers to process data in the form of text, images, or sound in a way that is inspired by the human brain. It can recognize complex patterns in these forms of data and produce accurate insights and predictions. It uses neural networks to achieve this. The input data is being passed to a hierarchical set of neurons which are capable of learning to recognize complex features necessary for producing the required result.

These neurons are stacked into various layers, causing an hierarchical structure, thereby forming deep neural networks. All these neurons except the ones in the output layer have non linear activation functions. The output neuron is modified based on a regression or classification task as having sigmoid activation or softmax activation.

The DNN model contains two passes for training - forward propagation and back propagation

Forward propagation : The input data is being passed to the model and each of the hierarchical neurons grabs a feature based on their weights and finally produces the output based on the input task.

Backpropagation : In this step, we use the loss function to update the parameters of the model using optimization algorithms such as gradient descent.

Deep Residual learning classical models use Deep learning techniques to train and perform inference on the dataset. The main advantage of using Deep Residual learning [8] comes from its architecture. Previously, as the depth of the neural network increases to obtain more trainable parameters, the gradient propagation from the final layer to the initial layer either vanishes causing a vanishing gradient problem or explodes causing an exploding gradient problem. To overcome this issue, Deep Residual networks come into play.

Deep Residual networks contain various skip connections between the neural network layers. During forward propagation, it moves through all the layers without skipping and produces the prediction. But during backpropagation, the gradients flow through the skip connections and thereby avoiding vanishing and exploding gradient problems while training.

Hybrid Quantum Classical Machine learning model

Classical models which use deep learning techniques are proven to be good feature extractors for the provided data. These models can be trained using various optimization techniques to extract the best features from the input data for which the prediction is being generated. On the other hand, quantum computing enables us to tap into the higher dimensional Hilbert space using various gate operations. By combining both the techniques, we can gain a lot of advantages in modeling.

The quantum model architecture comprises 4 sections - initial state preparation, transformation, trainable ansatz and measurement.

Initial State preparation - In this stage, the initial qubit state is being prepared. It can be in all 0 states or it can be in superposition of all the 2^n states for n-qubit circuits.

Transformation -  In this stage, the classical data is being transformed into the hilbert space. In this hybrid architecture, we use angle embedding to achieve this. The classical data from the resnet34 is being taken as the angle of rotation within the Hilbert space. The embedding can be done by either X, Y or Z rotation in the Hilbert space.

Trainable Ansatz - Here, we perform the training operation. The Ansatz contains a set of gates along with linear entanglement. The angles of these gates are being optimized to converge into the final solution.

Measurement - This is the final stage of the model where we convert the data from the hilbert space to the classical world.

Final Models for Various Hypotheses:

Hypotheses Model Accuracy (%)
Hypothesis - 1.1 : Early detection of dementia Resnet34 + QML
Qubits - 4
Depth - 4
Hypothesis - 1.2 : Early detection of Mild Cognitive Impairment (MCI) 16 images in a 2D plane preprocessed MRI
Hypothesis - 1.2 : Early detection of Mild Cognitive Impairment (MCI) Middle Images
Hypothesis - 2.1 : Drug Target Prediction for Alzheimer’s disease Hybrid QML
Model 4 - 140 input binary features
Angle Embedding -
Basic Entangler (with final to first CNOT)
Qubits - 4
Depth - 2
Hypothesis - 2.2 : Drug molecule generation for Alzheimer’s disease Generative RNN Train loss : 1.65

Model deployment in Website

A Flask backend based Web Application has been created to deploy the model so that it can be used for inference. The Web application is developed using Flask as backend and using HTML, CSS, Bootstrap for the frontend. The application is also containerized into a Docker container so that the requirements required to run the application can be installed on the container without affecting the server configuration.

A DockerFile has been created which uses the official Pytorch Linux environment to set up the Operating systems and necessary packages. The port 8080 inside the docker is also exposed to the outside world to access the application. It is also deployed in GCP AppEngine services as a serverless application so that GCP manages all the IT related infrastructure aspects.

Images of the Running Website

Loading Screen


Landing Page


Hypothesis 1 Page


Hypothesis 1.1 Prediction Page


Hypothesis - 1.2 Prediction Page


Hypothesis 2.1 Prediction Page


Hypothesis - 2.2 Prediction Page



  1. Life after a dementia diagnosis and dementia life expectancy. Age Space. (n.d.).

  2. What is alzheimer’s? Alzheimer’s Disease and Dementia. (n.d.).

  3. Rasmussen J, Langerman H. Alzheimer's Disease - Why We Need Early Diagnosis. Degener Neurol Neuromuscul Dis. 2019 Dec 24;9:123-130. doi: 10.2147/DNND.S228939. PMID: 31920420; PMCID: PMC6935598.

  4. McGleenon BM, Dynan KB, Passmore AP. Acetylcholinesterase inhibitors in Alzheimer's disease. Br J Clin Pharmacol. 1999 Oct;48(4):471-80. doi: 10.1046/j.1365-2125.1999.00026.x. PMID: 10583015; PMCID: PMC2014378.

  5. Dubey, S. (2019, December 26). Alzheimer’s dataset ( 4 class of images). Kaggle.

  6. Alzheimer’s disease neuroimaging initiative. ADNI. (n.d.).

  7. ChemBL Database. EMBL-EBI homepage. (n.d.).

  8. He, K., Zhang, X., Ren, S., & Sun , J. (2015). Deep Residual Learning for Image Recognition. arXiv.