We propose a supervised learning approach to statistically quantify the impact of an extreme event on vulnerable communities using publicly available panel data directly reflective of the different dimensions and manifestations of social hardship. These manifestations include suicides, substance abuse, excess mortality, unemployment, and others. Our modified treatment-effect model allows counterfactual baseline conditions to be posited for each manifestation from which an aggregated quantitative multi-faceted measure of social hardship can be determined. The developed statistical methodology should be greatly beneficial to policymakers who must allocate scarce resources to mitigate social hardship. Our work represents a distinct alternative to the established approach of assessing social vulnerabilities of communities subject to extraordinary events that rely on composite indices (such as Social Vulnerability Index SoVI) based on published census data. We illustrate applicability of our approach using annual and monthly panel data from 2012-2018 encompassing the 2017 Hurricane Maria event across various municipalities in Puerto Rico. Our statistical modeling methodology stands apart since (i) it explicitly and more realistically captures the effect of different manifestations of the actual event, (ii) it is flexible enough to accommodate individual preferences of various stakeholders in how they assign importance to multiple manifestations of social hardship.