The PARIS Demo

This page demonstrates how you can leverage medical questionnaire data in MEDomics to draw insights.

The original data is not available. We have generated a synthetic version using the synthpop package, and it will be made available on Zenodo once we receive authorization from the PARIS team (expected in mid-2026).

About the Dataset

Patient-Reported Indicators Surveys (PARIS) is a combination of PROM (Patient-Reported Outcome Measures) data, which focuses on patient health status, and PREM (Patient-Reported Experience Measures) data, which captures the patient's experience with care. It represents a questionnaire completed by patients to provide a complementary view of healthcare quality, enabling organizations to track progress, identify areas for improvement, and personalize care. The data's columns/questions are illustrated in the table below.

Goal

This POC aims to show how the PARIS data can be exploited using the MEDomics platform. We will explore how Superset can be used to both visualize and curate the data. Moreover, we will review several MEDomics modules to demonstrate the potential of each one in data processing and modelling, with the ultimate goal of predicting a pre-selected variable, in this case, the patient's emotional distress.

Initially, the PARIS dataset contained about 200 columns (questions), and to simplify this demo, we manually chose those we believe are linked to the clinical issue of mental distress.

Column Name	Question	Possbile Answers
EnergeticVigorous2	I felt energetic and vigorous.	All the time Most of the time More than half the time Less than half the time Occasionally Never
DailyLifeInterests2	My daily life has been filled with things that interest me.	All the time Most of the time More than half the time Less than half the time Occasionally Never
SleepRested2	I woke up feeling rested and refreshed.	All the time Most of the time More than half the time Less than half the time Occasionally Never
Fatigue7	Over the past 7 days, how would you rate your average level of fatigue?	None Mild Moderate Intense Very intense
ActivitiesPain7	Over the past 7 days, to what extent has pain interfered with your daily activities?	Not at all A little Moderately A lot Extremely
Pain7	Over the past 7 days, how would you rate your average pain level?	0. 0- No pain 1 2 3 4 5 6 7 8 9 10- The worst pain possible
SocialRoles	Overall, how do you feel you are fulfilling your usual activities with others and your role in society (whether at home, at work, in your immediate environment, as well as your responsibilities as a parent, child, partner/spouse, employee, friend, etc.)?	Excellent Very good Good Mediocre Poor
PhysicalActivities	To what extent are you able to perform daily physical activities such as walking, climbing stairs, carrying shopping bags, or moving a chair?	Completely Almost completely Moderately A little Not at all
Age	Age	44 years old or younger Between 45 and 49 years old Between 50 and 54 years old Between 55 and 59 years old Between 60 and 64 years old Between 65 and 69 years old Between 70 and 74 years old Between 75 and 79 years old Between 80 and 84 years old 85 years old or older 97. I prefer not to answer
Sex	Sex	Female Male Non-binary Other I prefer not to answer
NutritiousMeals12	Have enough money to buy nutritious meals?	Always Often Sometimes Rarely Never
RentMortgage12	Do you have enough money to pay your rent or mortgage?	Always Often Sometimes Rarely Never
MonthlyBills12	Do you have enough money to pay other monthly expenses, such as your electricity, heating, and phone bills?	Always Often Sometimes Rarely Never
DiscussionHealthcareProfessionals	Do you discuss with your healthcare professionals what is most important to you in managing your health and well-being?	Not at all To some extent Most of the time Always Not applicable
HealthcareInvolvement	Are you as involved as you would like to be in decisions about your care?	Not at all To some extent Most of the time Always Not applicable
HealthcareConsideration	In the context of your care, are you treated as a whole person and not reduced to a disease or health problem?	Not at all To some extent Most of the time Always Not applicable
ComplexityHealthIssues	Most health issues are too complex for me to understand.	Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree
target	In the past 7 days, how often have you been bothered by emotional problems such as feeling anxious, depressed, or irritable?	Never Rarely Sometimes Often Always Binarisation: [1, 2] values are converted 0 i.e. no mental distress, whereas [3, 4, 5] values are converted to 1, i.e. mental distress.

Steps

Here are the steps followed in this demo:

1

Superset

Create customizable dashboards to gain a deeper understanding of your dataset, and utilize the embedded SQL Lab to prepare your data for the next steps.

2

Exploratory Module

Use different tools, such as sweetViz, to understand the underlying relationship between your data's variables and potentially delete redundant ones.

3

Input Module

Multiple tools can be exploited in the Input Module, such as the Create Holdout Set Tool, to partition data into training and holdout sets.

4

Learning Module

The Learning Module represents the main step of the demo. It will be utilized to test multiple machine learning algorithms for predicting the emotional state variable, select the best-performing one, and train and save a final model.

5

Evaluation Module

In this module, we will utilize the saved machine learning model to make predictions on the holdout set and try to interpret and explain the model's choices.

6

Application Module

This final step is similar to model deployment, where we will utilize the saved model from the Learning Module to generate predictions on new input data.