Models for Airborne Infectious Diseases: Effective & Novel (MAIDEN)

Models for Airborne Infectious Diseases: Effective & Novel (MAIDEN) are needed to deal with future outbreaks and pandemics. These models need to be accurate or effective to deal with the complex spatially distributed dynamics of such emergencies. MAIDEN follows a first-principles approach to develop complete models of aerosol transmission.

Aleck H Alexopoulos, PhD - Primary Investigator

Current epidemiological models are based on SIR variants. These models are valid when transmission is predominantly via direct contact but completely invalid when the main mode of transmission is by aerosols. Data-based models have also been proposed but these require lengthy periods of time to accumulate sufficient data for learning.
More specifically, for covid-19, SIR models presented uncertainty in predictions on the order of +-50% over a two-week period. Those that appeared to follow the disease dynamics updated model parameters on a daily basis (and thus were not predictive). One characteristic failure of SIR-models is that they predict abrupt drop-off in infection rates when in fact mostly there was a long decrease in the number of new cases. This aspect of the disease is completely unpredictable by SIR models and the reason for this failure is that SARS CoV-2 is in fact transmitted by aerosols.
Aerosol transmission involves not just people but also closed spaces. An infected person can enter a small store or a bus, exhale infected air for a period of time, leave and then a second person enters the room later and be exposed to CoV-2. 
Models capable of aerosol transmission do not exist Until MAIDEN!

MAIDEN can provide
1. accurate dynamics of an airborne disease where SIR models fail.
2. accurate predictions of the effects of measures.
3. assistance to regulators to determine ventilation needs of new spaces
MAIDEN is helpful to Scientists (Epidemiologists and others) as well as Regulators and City Planners
Presently I am tracking the problem side, i.e., the failure of SIR-models and the confusion and disagreement amongst experts about which measures can be effective.

Our goal is to develop these algorithms and implement them first in a simulated neighborhood and then in more realistic real-world case studies.
MAIDEN will result in white paper reports, publications, and an open-source model available to all. Benefits will be to:
1. Scientific community e.g. epidemiologists and anyone dealing with pandemic modelling and intervention measures
2. City planners, regulators etc as a tool to assist in developing guidelines and regulations e.g. for ventilation fin terms of air renewal per unit time or per unit time and per person.
3. Scientific community for the development of deeper models of infection and transmission

Our project MAIDEN will provide open-source models which will enable the Scientific community, e.g. epidemiologists, to perform much more accurate simulations and predictions of a pandemic and better evaluate possible intervention measures. With this information much more effective reactions and control of future pandemics will be possible leading to shorter duration of the pandemic, fewer deaths, and much less economic slowdown.
Evidence: hundreds of papers during covid attempting to do just this and failing. e.g.
John P.A. Ioannidis et al. 2020 "Forecasting for covid-19 has failed"
https://www.ncbi.nlm.nih.gov/p...

First, by the end of the first year, we aim to perform simulations on a small virtual neighborhood and then a larger real area in order to evolve the models to a degree where they are of practical use to the scientists such as epidemiologists. Further refinements over the second year together with input from external users will enable an interface through which non-expert users can perform simulations.
Scale is really achieved by providing the open-source code after the end of the first year and then a more user-friendly version with a front-end interface by the end of the second year and refinements during the thrid year based on user feedback.

First year indicators.
1. Simulations of virtual neighborhoods. KPI's will be to generate diseases dynamics and responses similar to those seen during the covid-19 pandemic. KPI's could be identification of four to five key parameters that determine clearly and independently the key aspects of aerosol transmission - controlled pandemic dynamics. These parameters would be physically meaningful parameters such as Ro and K.
2. Simulations in a real neighborhood determine of model robustness and sensitivity. KPIs would be sensitivity indexes.

One is financial. Although not much funding is necessary for this project. Less than $100k in 1yr to provide the first open-source code.
The second is academic ego. After the very bad performance of the SIR-models many of the Scientists working in this area have become "sensitive" to criticism and, worse, averse to new ideas.This is one reason I have not yet attempted to publish this work. Independent funding and providing an open-source code is the way past this problem. We do agree that a continuation of 2yrs just to disseminate, educate and promote the solution will probably be necessary.

CERTH
Center for Research & Technology Hellas, Greece
www.certh.gr

I am fascinated by the challenge and have been deeply involved since the beginning of the covid-19 pandemic.
Looking for support to develop complex algorithms that we then intend to provide for free as open-source code is not an easy task.

The goal is to provide a solution that can help deal with future emergencies.
We have several connections to the Medical Community but would benefit from an advisor from a relevant International Health Organization.