The Team.

The Data Jones Team is composed by a series of actors coming from different realities and with mixed backgrounds. They met with the idea to combine their capabilities to widen analyse the chance to find insightful data mapping by leveraging the huge amount of data already available. From EXOM Group, Luigi Visani. He is a medical doctor by training and an expert of clinical research. For years through Exom Group, a Contract Research Organization (CRO) he has organized and analyzed dozens of clinical trials with many drugs and in many therapeutic areas, including the recent Covid-19 pandemic. Data science and advanced analytics are also one of his areas of expertise and actual experience. From Machine Learning Reply, Giorgia Fortuna, Elena Facco, Francesco Fontan e Nicola Bottone. Machine Learning Reply delivers a full package of AI-related services: data strategy consulting, innovation management consulting and AI solutions development. The Machine Learning Reply’s people are experienced in deep learning, computer vision, NLP and predictive modeling. Machine Learning features Certified Data Engineers on the Google Cloud Platform and is part of the Nvidia Partner Network. Nicola has been working for Reply since 2012 and is dedicated to defining new business models and evolutions in the offer, as well as structuring the products and services exploiting the new technologies and potential offered by the IoT/IoMT (Internet of Medical Things), the Cloud and ML/AI. Giorgia has been working in Reply for the past three years and she is dedicated in shaping end-to-end ML solutions for Reply’s customers giving support in the design of the solution and in the delivery of it. Elena is a Data Scientist with expertise in machine learning and biophysics. She has been working in Reply for the past three years. Francesco is a Data Scientist who loves modeling problems in mathematical terms. He has some expertise in Deep Learning ranging from Computer Vision to NLP. From Healthy Reply, Nalesso Yuri e Di Nuccio Alfredo. Yuri has been working for Healthy Reply since July 2019 as a fullstack developer. His background is of human data analyst and he is currently attending a master in artificial intelligence in Politecnico di Torino. Alfredo has been working for Healthy Reply since November 2020 as a full stack developer. He has a background in data science and during his university career he worked on various projects in the field of machine learning applied to medicine.

What we know so far.

Several projects and innovation activities have been initiated in the last months with the same intent, to fight COVID-19 pandemic. COVID-19 Datasets popped up everywhere around the world and, right from the start, the idea to use recent technological tools, Artificial Intelligence (AI) Tools in particular, has been explored. The projects already running are focused on a very different angle of COVID-19: calculate prognostic risk, epidemiology, early detection, diffusion and even a spatiotemporal epidemic model to quantify the effects of contact tracing.

Stratify and conquer.

Over the last year the entire global population has been affected by COVID-19 virus and we have a large data-set with huge amounts of data from different countries and from people of different ethnicity, age, and, generally speaking, with different clinical and socio-sanitary conditions. In the last years a process of stratification of patients affected by Chronic Disease has started. The idea was to identify a global set of parameters (e.g. clinical, social, geographical) concurring to their pathological condition. This process brought many advantages that we aim at replicating in the Covid-19 context. The Team, in fact,aims at using the available data to define a “class of risk” and to identify the best approach to patients’ pathology over time. This approach will support both patients and healthcare professionals. Patients will undergo a standard but also personalized clinical and therapeutic pathway of validated interventions. The healthcare professional gets support in defining a personalized clinical path based on solid and accurate data and then building on it. Starting from this idea, already running for Chronic Disease, The Data Jones Team would like to build a concrete stratification mechanism for patients affected by COVID-19. The availability of appropriate Risk Indicators and Classes, through the stratification, will allow healthcare professionals to standardize procedures and to create a sort of guideline reducing the time for the clinical-prognostic assessment of the patient. Moreover, The Data Jones Project will contribute to making the patient become more aware of his clinical condition.

Generalize and reproduce.

The team will approach the challenge of defining the above stratification in a systematic and flexible way to ensure reproducibility but also to build a methodology that professionals can apply to other types of data. The advantages of making this is that many of the techniques that the team will use can be generalized to fit other problems. This will lead to the creation of a platform where both professionals and patients can find explainable results based on the uploaded data and the pathology to trace.

Approach and Technology.

Specifically, the process will be focused in identifying a methodology or an approach starting from a well known background on Chronic Disease and to generalize it for COVID firstly. The main points identifying the strategy to build the methodology can be summarized in:

Check trends, patterns and anomalies on the data provided. The idea is to even understand if the patients are entering a risk zone and so to support the healthcare professional to act on time and proactively.
Correlation identification and then usage of explainable models. Here the main goal is to understand, between the variables analyzed, the most impacting on the health condition of the patients. Moreover, it will be interesting to study how these most impacting variables are correlated between them and the patient’s health condition.
Build the whole data history aggregating all the information retrieved and then comparing these data with a dataset coming from people not affected by the virus.
Generalisation of the methodology for COVID variation and thus different pathologies. Data set may vary from different hospitals or healthcare providers and so it's necessary to understand how the model needs to be updated.
GUI for professionals to facilitate the data usage and understanding and smooth their mission.

Benefits where you are.

Using this approach and methodology, the professionals have the chance to determine a set of guidelines to approach the patient’s health condition. The monitoring and predictive evolution of the patient’s healthcare condition will lead in minimizing the action time and prevent the worsening of the patient’s clinical condition. The healthcare professional, then, will be able to personalize the treatment based on the actual condition and predicted one. Moreover, this approach will result in better/optimal resources’ utilization (e.g. intensive care machines occupations, drugs request, healthcare professionals shifts). The data collected will be used not only by the healthcare professional but by the patient too. A patient who is aware of his or her clinical condition and the potential risks he or she faces, improves compliance with treatments and cooperation with health care personnel.