COVID-19 Detection Tools Raise Issues About Privacy in Public Health-Related Monitoring

Screenshot of the disease visualization tool illustrating how infection hotspots have developed and resolved over time

Effectively combating the COVID-19 pandemic depends on the detection and containment of concentrated areas of infection as well as intervention efforts that prevent widespread community transmission. To that end, the U.S. government is adopting technologies to support the public health response to the COVID-19 pandemic and to increase national preparedness for infectious diseases more broadly. A key component of disease outbreak management is timely and targeted information on the incident location, affected persons, infection transmission time, and other associated disease risk factors. 

Public health activities such as COVID-19 case identification, contract tracing, and vaccination implementation performed across an entire jurisdiction can be a daunting task, especially with limited resources. In response, researchers at the Centers for Disease Control and Prevention have developed a tool for use by public health officials that enables hotspot detection of COVID-19 outbreaks at the granular level. That is, pinpointing specific residential areas with critical COVID-19 concern. Using the tool – whose documentation was released June 30, 2022 – public health authorities can tailor their efforts to a particular neighborhood or congregate living facility instead of a whole zip code area. 

As pictured above, this tool is an interactive, real-time, street-level visualization of when and where COVID-19 cases are developing and resolving. The algorithmically-defined alerts generated can trigger government responses to public health threats and disease outbreaks in a timely and targeted fashion. In addition to displaying COVID-19 hotspots, the automated tool generates contextually rich illustrations of patterns of change and notable trends across the course of the pandemic. For example, it analyzes inputted data to create a representation of the distribution of sociodemographic and socioeconomic characteristics among COVID-19 outcomes in specific geographic areas. Social determinants like race and income are factors that collectively influence the health of an individual and inequities in the provision of societal resources can have disproportionate impacts on certain populations/people groups . On that account, the sociodemographic and socioeconomic characteristics of COVID-19 hotspots provided serve as critical contextual knowledge that might help to inform the types of public health measures developed by government officials. The tool provides government officials with predictive power when it comes to assessing future pandemic-related risks and needs. With these capabilities, it has the potential to become distinctly influential in high-impact decisions made by public health officials. However, in the event of a malfunction, offloading a considerable amount of decision making to the tool could lead stakeholders astray and cost them precious time and resources. 

Given the prolonged nature of the COVID-19 pandemic and our uncertainty with what the future holds, the tool represents a step toward adapting to the challenges of our new reality. In fact, COVID-19 could very well become a standard seasonal disease just like the flu. The CDC anticipates that the tool will not only prove useful for the public health department with respect to the COVID-19 pandemic but eventually to other public health threats including chronic diseases as well. 

Source data including patient demographics (age, race, ethnicity, sex), street address of residence, test results, and test date from regional public health systems informs this tool. Sociodemographic and socioeconomic context was generated from the Census Bureau’s decennial census and the American Community Survey’s data and details on the prevalence of chronic diseases was derived from county-level data from the Behavioral Risk Factor Surveillance System. While rapid and informative, the tool is largely restricted by publicly available sources of data that do not provide information at smaller units of geography. Inconsistencies in government reporting of statistics and the inaccessibility of useful datasets further intensifies the challenges in quality and consistency of data collection that are used to inform these systems. 

While imperfect datasets and selection bias raise questions about the accuracy of this tool, it is worth noting that tracking infection at this intended level of granularity presents privacy and confidentiality concerns. This highlights how the utility of this tool must be balanced with appropriate accountability. During the Ebola outbreaks in West Africa in 2014 to 2016, personal phone data was used to monitor individual’s mobility and behaviors. There is potential for COVID-19 tools to also tend toward the questionable usage of personal data in monitoring individual mobility, social-distancing measures, and quarantine adherence.

Enabling the creation and legitimization of government surveillance tools during a period of national emergency can present both benefits and risks. Although useful in the present time, these tools are likely to persist after the pandemic for non-emergency purposes. State governments could, for instance, duplicate and deploy their own versions of the tool to track other types of health data that is of interest to them. Instead of monitoring COVID cases, the tool’s automated capabilities might be leveraged to track, for instance, pregnancy cases. In a state with anti-abortion laws, the tool’s use would have serious implications for women. Detection tools open up potential avenues for government analyses of health that may drastically improve existing measures. But in doing so, they raise serious concerns over privacy rights.

As tools like these continue to evolve over the course of the pandemic, the ongoing collection of data from private citizens merits rigorous investigation and routine checks of these digital technologies. Increasingly accurate epidemiological models that have the power to predict future outcomes require legal and ethical considerations, greater transparency, and meaningful public participation, knowledge, and scrutiny. 

Algorithms for Modeling, Mapping, and Mitigating Wildfire Risk

Government agencies have adopted various algorithms to combat the wildfires that have ravaged much of California and the Pacific Northwest this past summer. Just like COVID-related government algorithms, wildfire-related algorithms present a variety of approaches to help understand and address a widespread issue.  

Many of the wildfire-related government algorithms that we found at Algorithm Tips aim to identify and model wildfires in real-time. For example, Sonoma County, California, incorporated a computer vision-based tool called ALERTWildfire for detecting fires in preparation for this year’s wildfire season. NASA’s IMERG algorithm uses satellite data to estimate precipitation over the majority of the Earth’s surface, and in turn, inform models of fire conditions. Similarly, the National Integrated Drought Information System wildfire conditions map visualizes active wildfires alongside current drought conditions across the country. 

Other algorithms make predictions about wildfire risk. For instance, the Oregon Wildfire Risk Explorer creates custom maps, detailed summaries, and risk reports based on public wildfire data. At a national scale, the National Fire Protection Association (NFPA) has been piloting a digital community risk assessment tool that similarly helps visualize fire and accident risks within a given community through customized maps, charts, and graphs based on various data related to geography, economics, demographics, and infrastructure.

These wildfire-related algorithms are important given the serious and widespread threats that wildfires pose. The 2021 Dixie Fire burned nearly a million acres, for example. In 2019, wildfires caused an estimated $4.5 billion in damages in California and Alaska, and since 2000, 15 forest fires in the U.S. have caused at least $1 billion in damages each. Recent decades have seen an increase in forest fire activity over the western U.S. and Alaska, and as climate change continues to exacerbate natural hazards like wildfires, these algorithms will only continue to grow in importance. 

Risk modeling algorithms are especially important as they inform other decision-making processes. The aforementioned algorithm from Oregon informs professional planners as they update Community Wildfire Protection Plans, for example. Similarly, Savannah, Georgia, one of the communities in the NFPA pilot program, explained how the NFPA dashboard will help it make data-informed decisions about local fire prevention efforts. Furthermore, as in the case of flooding, fire risk assessments also inform insurance policies

As journalists cover the development of firefighting efforts, they can investigate the new technologies, such as computer vision and artificial intelligence, that government agencies have adopted to combat wildfires. Journalists can also investigate the various fire-related risk assessments guiding policy and insurance decisions as they are made in response to active wildfires. 

Eco-Scoring: How EPEAT ratings encourage a more sustainable tech future

The Electronic Product Environmental Assessment Tool (EPEAT) has come a long way since being introduced in 2006. Managed by the non-profit organization Green Electronics Council and originally commissioned by the US Environmental Protection Agency (EPA), the EPEAT algorithm helps governments, institutions, consumers, and other purchasers evaluate the effect of a product on the environment. During a time when there are increasingly more electronic products being developed, many of which can have undesired consequences on the environment, tools such as the EPEAT are crucial.

The EPEAT can help both companies and buyers contribute to sustainability through its comprehensive ranking system. As an eco-label for the IT sector, it is specifically aimed at enabling consumers to make conscientious decisions when buying electronics, which in turn pressures manufactures to create sustainable products.

What Are EPEAT Ratings?

The EPEAT labeling system is based on various sustainability categories, such as use of post consumer recycled plastic; responsible end-of-life management; substance management; design for repair, reuse and recycling; and so on. These sustainability categories are applied to product categories including: Computers and Displays, Imaging Equipment, Mobile Phones, Photovoltaic Modules and Inverters (PVMI), Televisions, and Servers.

The EPEAT is already impressive with its environmental criteria for different products, and its wide range of performance categories creates a well-rounded picture for buyers wanting to be informed. It could likely be improved even further by expanding the environmental criteria to involve a range so consumers can see just how extensively a product meets each of the requirements.

Here is a look at the EPEAT’s environmental criteria for computers:

If a device meets all the required criteria and a certain percentage of optional criteria, it is awarded one of three possible ratings:

  • EPEAT Bronze: A device meets all required criteria.
  • EPEAT Silver: A device meets all required criteria and at least 50% of the optional criteria
  • EPEAT Gold: A device meets all required criteria and at least 75% of the optional criteria.

The EPEAT is especially valuable to three groups: purchasers, manufacturers, and environmental advocates. Purchasers will find the verified and credible tool easy to use, providing a single source to identify qualified products without the need to perform technical analysis. Manufacturers are benefited by the harmonized standard and no delay in time to market, while environmental advocates are presented with verified manufacturer claims and an impactful tool for environmental leadership. The EPEAT-approved Conformity Assurance Bodies (CABs), which are third party testing and certification organizations, provide third party verification of manufacturing claims by conducting a set of comprehensive review processes. These processes involve registration materials and the analysis of manufacturer responses to verification questions.

Image: Stakeholder Categories, Global Electronics Council

The EPEAT is a good example of how public sector demand can lead to new product markets, as well as stimulate eco-innovation in the private sector. Consumers can feel confident in the criteria, which were born out of multi-year stakeholder processes and continue to undergo improvements. The stakeholder processes involve hundreds of representatives from governmental, environmental, research, and manufacturing sectors (see figure inset). Public and private bulk consumers also contribute to the processes, and EPEAT partner organizations provide product rating information in their catalogs and train sales staff to support EPEAT purchasing requirements. All of this leads to raised consumer awareness and creates far more transparent products.

EPEAT-registration services are provided by multiple international organizations, and this spans across North America, South America, Europe, Asia, and Australia. Each one of the organizations has auditors that are qualified to evaluate the conformance claims of electronics manufacturers and supplies. The EPEAT Policy Manual states that auditors become qualified to perform conformity assurance of product categories by attending Initial EPEAT Auditor Training, which they then have to follow up by passing the Initial EPEAT Auditor Exam for each product category. Each product category requires separate training and exams.

Towards Accountability

The impact of the EPEAT was felt immediately following its release in 2006 with 60 products from three separate PC and Display manufacturers. In 2007, President George W. Bush issued Executive Order 1342 requiring all US Federal agencies looking to purchase computer systems to use EPEAT. This was then renewed in 2009 by President Obama, which preceded Executive Order 13514 requiring all federal agencies to purchase at least 95 percent of their electronics based on EPEAT status when applicable.

In 2012, the EPEAT’s ability to hold private companies accountable to the public was demonstrated when Apple’s 2012 Retina Display Macbook Pro failed to meet the EPEAT’s criteria for disassembly. That laptop was subsequently named by as “the least repairable, least recyclable computer encountered in more than a decade of disassembling electronics.” Greenpeace followed by denouncing Apple and saying the company was “greenwashing,” or spinning its PR to appear as an environmentally friendly company.

Apple responded by pulling all of the company’s 39 certified desktop computers, monitors, and laptops from the registry. During this controversy, the City of San Francisco announced that it would no longer be buying Macs following Apple’s withdrawal, and that none of the company’s products would be allowed in the city’s 50 agencies. However, this did not last long as Apple realized it would have trouble surviving without a positive EPEAT rating, so it rejoined and relisted all of its qualifying products on the registry.

As for algorithms and tools that have a positive societal impact, the EPEAT is arguably one of the most important in this regard. Without such a tool, consumers would have a difficult time being informed about what goes into the countless technological products they use in their daily lives. It creates much needed standards for electronic products, which is important for consumers all around the world looking to make informed and sustainable purchasing decisions. Given how electronic products are often produced under harmful environmental conditions, and in a way that makes them hard to be recycled, EPEAT incentivizes organizations to develop more ecologically responsible products.

Improving EPEAT

Though largely a force for good, the EPEAT is not without its flaws and fair share of criticisms. One of the more concerning questions surrounding the EPEAT is whether or not the criteria are applied equally across all organizations. The labeling registry has been criticized by organizations like Greenpeace for “caving in” to pressure from leading manufacturers, especially Apple.

Back in 2012, the EPEAT undertook a thorough review of five separate “ultra-thin” notebooks from Apple, Lenovo, Samsung, and Toshiba, which had been proven to be difficult to recycle. The registry eventually approved them, which confused many consumers and businesses looking to get clear direction from the EPEAT. One of the reasons for the approval was that the new products could indeed be disassembled. However, this required the removal of components from the gadgets, which most consumers would be unwilling to do given product warranties. Events such as this undermine the EPEAT’s credibility and begin to hurt the public’s trust in such systems.

Despite these valid criticisms, it is more important than ever for technology companies to be held accountable in various ways, especially when it comes to environmental sustainability. A comprehensive and trustworthy labeling system like the EPEAT moves us in that direction as it pushes industry leaders, producers, and suppliers to meet the highest standards of environmental and social sustainability.

The EPEAT should continue to improve its rating criteria and assure consumers they can trust the process. If it takes more steps in this direction, the EPEAT could prove to be a glowing example of how the government can make a difference with such technologies, as it helps create a far more transparent private sector, which is key as we progress further as a technology-driven world.

Medicare algorithms will estimate patient race and ethnicity

In August 2021, the Centers for Medicare & Medicaid Services (CMS) authorized the implementation of an algorithm that tries to predict the racial and ethnic identities of patients. While this may allow the CMS to identify racial and ethnic groups that suffer from lacking health equity, its use can be controversial, particularly when the accuracy of the system fails.

The Medicare Bayesian Improved Surname Geocoding (MBISG) model was developed by CMS with the help of a contractor to estimate the probability of Medicare Advantage Plan beneficiaries belonging to each of six racial/ethnic groups. The MBISG model calculates these estimates by using Medicare administrative data, first and surname matching, and geocoded residential addresses linked to the 2010 U.S. Census.

Though the model results in high concordance levels between its estimates and self-reported race and ethnicity by certain individuals, the CMS recognizes the model’s decreased accuracy for those who self-identify as American Indian, Alaskan Native, or multiracial. CMS also recognizes the potential for bias being introduced into the model due to factors such as this decreased level of accuracy, which can be problematic since the end goal of the MBISG model is to evaluate the quality of care being received by each racial/ethnic group. 

The MBISG model arose as an effort to mitigate the lack of racial and ethnic data for Medicare beneficiaries, which the CMS uses to track health equity gaps among different groups. The CMS recognizes that self-reported race and ethnicity data is preferred, but not always available or accurate due to limited race/ethnicity options in the past as well as reluctance to disclose this information among certain people. The MBISG estimations are meant to account for situations in which individuals might have identified with a racial or ethnic group that was not available as an option for them when self-reporting, which was often the case prior to 1980 when the only options available were “White,” “Black,” and “Other.” 

Bloomberg Law published an article covering the MBISG model on August 5, 2021. The Bloomberg article describes CMS race/ethnicity estimation algorithms including the MBISG, points out their flaws, and encourages the use of caution when employing these algorithms due to bias concerns. The focus is mainly on why this technology is flawed and its usage concerning. However, the article also emphasizes the purpose of these algorithms and how the CMS intends to use them as part of the process of understanding health inequities among different racial and ethnic groups.

A key point in the article regards how medical groups, such as the Association of American Medical Colleges perceive proposed estimation algorithms such as the MBISG model. They are generally concerned with the accuracy of the data provided by such algorithms and push for the CMS to invest in improved patient data collection methods. However, CMS states that despite past efforts to improve data collection and quality, Medicare administrative data continues to present varying accuracy when it comes to identifying patients belonging to different racial and ethnic groups.

The Bloomberg article suggests that, in theory, the implementation of the MBISG model and similar algorithms can have a positive impact because they provide a more accurate representation of the demographic makeup of Medicare beneficiaries. This allows CMS to detect which groups might require additional care if the care they are currently receiving is inadequate. However, it is important that both Medicare and CMS are transparent with patients and other stakeholders regarding these algorithms due to their use of personal information and their potential to impact decision-making in healthcare

Journalists can continue researching these estimation algorithms in healthcare — and their potentially controversial implications — by exploring the transparency aspect of the situation, discussing how CMS and Medicare are planning on disclosing information regarding these algorithms to stakeholders including doctors and patients. It would also be interesting to look into how Medicare plans on improving their current patient data collection methods since this is an important process and, if ever sufficiently improved, data collection could remove the need for these types of estimation algorithms in healthcare. 

​​Scoring Environmental Inequality: How California is paving the way for the U.S.

Last month, California released a new draft of the California Communities Environmental Health Screening Tool (CalEnviroScreen), a web-based tool for identifying communities that are disproportionately burdened and endangered by pollution. Using environmental, health, and socioeconomic data from state and federal sources, CalEnviroScreen scores the pollution burden of every census tract in the state and presents these evaluations in an interactive color-coded map that allows users to explore and compare different communities. 

Reproduction of the CalEnviroScreen map, showing the disparity between San Francisco and Oakland

In identifying communities suffering from the cumulative impacts of multiple pollutants and people who are vulnerable to the effects of pollution, CalEnviroScreen aims to address racial and economic inequalities throughout the state. Academics, activists, and nonprofit organizations alike have acknowledged how low-income communities and people of color are disproportionately impacted by and vulnerable to the negative effects of pollution. Studies show that although white Americans’ consumption disproportionately causes air pollution, Black and Hispanic Americans are disproportionately impacted by it. Across the country, people of color and people in poverty are disproportionately exposed to facilities that emit hazardous particulate matter. Activists and nonprofit organizations have also brought attention to the toll that diseases like asthma have on communities with larger African American populations or higher poverty rates, and they have advocated for government policies to combat environmental injustices that contribute to these health disparities. 

To try to combat these inequalities, the California Environmental Protection Agency (CalEPA) uses CalEnviroScreen to inform resource allocation and policy decisions, such as the administration of environmental justice grants and enforcement of environmental laws. Ultimately, this tool serves to uphold CalEPA’s goal to ensure the fair development of laws and regulations that affect every community’s natural surroundings regardless of race, color, national origin, or income. Other state agencies also use CalEnviroScreen for a variety of programs, including those related to transportation and emissions, sustainable agriculture, and land preservation projects that reduce greenhouse gas emissions. 

CalEnviroScreen not only informs policy that affects nearly 40 million residents in California, but it could also impact the development of a similar tool at a national level. Like CalEPA, the U.S. Environmental Protection Agency’s goal is to “provide an environment where all people enjoy the same degree of protection from environmental and health hazards.” President Biden specifically included the development of a data-driven Climate and Environmental Justice Screening Tool as part of his campaign platform, and earlier this year, the White House began efforts to develop such a tool for identifying disadvantaged communities and informing equitable decision making across the federal government based on the EPA’s existing Environmental Justice Screening and Mapping Tool (EJSCREEN). The EPA began developing EJSCREEN in 2010 to highlight places to potentially focus environmental justice efforts, but unlike CalEnviroScreen, EJSCREEN and its underlying data are currently not used as a guidance for government agencies. Climate activists and the Center for American Progress have advocated for CalEnviroScreen to be used as a model for a national tool to address environmental justice concerns across the country.

With the federal government expressing interest in developing a similar environmental justice tool at the national level, and with climate change exacerbating existing racial and economic inequities, more debate and discussion on the issue can be expected. Questions about the fairness, representativeness, and effectiveness of these scoring systems remain open as these models are developed and adopted. Journalists can research the development of CalEnviroScreen since its first iteration in 2013, and they can monitor as CalEPA and other California agencies begin adopting the most recent version, draft 4.0. While CalEnviroScreen is perhaps the most notable model of a state-level environmental justice mapping and screening tool, journalists can also research similar efforts in other states (e.g. the Washington Environmental Health Disparities Map). At a federal level, journalists can learn more about EJSCREEN, which began in 2010,  and follow along with the national development of a Climate and Environmental Justice Screening Tool that began in January of this year. Those interested in these algorithms can also research the various studies (dating back to as early as the 1980s) that investigate connections between environmental injustices and racial and economic inequities. Looking to the future, journalists can research the effects that academics, activists, and other professionals expect climate change to have on environmental issues like air pollution and beyond. 

New algorithms to score candidates for lifesaving organ donations

COVID-19 has been in the spotlight not only in the news, but also in how technologies are being developed to evaluate and control the spread of the virus. For example, various state guidelines outline how ventilators, personal protective equipment, and COVID-19 vaccines should be prioritized in hospitals, ambulances, and communities, respectively. But while the pandemic illuminated how government algorithms shape the way healthcare resources are allocated throughout the country, government algorithms will continue to inform urgent healthcare decisions beyond the pandemic. And one of the services impacted in this is the allocation of organs for transplant.

A new score-based framework is updating the Organ Procurement and Transplant Network (OPTN), the national system for distributing lung, liver, kidney and other lifesaving organ donations. Individual organ systems have been transitioning to this new framework since the OPTN Board of Directors approved it in 2018: lung allocation was the first to be updated in January 2019, and liver, kidney, and pancreas allocations were updated last month. Rather than reviewing transplant candidates in ranked classifications and within fixed areas, these new algorithms continually calculate composite scores for candidates that weigh factors related to medical urgency, placement efficiency, outcomes, and patient access. A higher score puts a patient higher on the waitlist, and in turn, more likely to receive an organ transplant. This framework is supposed to be more equitable and adaptable to future changes, but as seen in the recent pushback against new kidney policies in particular, critics have argued that this change will increase wait times and give differential treatment to patients in densely populated regions. 

Organ transplantation is the leading form of treatment for patients with severe organ failure. There were over 32,000 organ transplants in 2019, and an average of 95 transplants now take place in the U.S. everyday. Unlike other life saving transplants (like those involving blood or bone marrow), most organ donations come from deceased donors. Unfortunately, there are not enough donations to meet organ transplant needs across the country: In 2020, about 110,000 people remained on national waiting lists, and currently, there are over 120,000 people in need of a life saving transplant. Someone is added to the national transplant waiting list every nine minutes, and over 20 people die waiting for an organ donation each day. 

Organ allocation systems not only determine who receives scarce organ donations but also what that medical care looks like. In addition to affecting wait times, allocation systems take into account compatibility between donors and patients, which affects the likelihood of transplant success. Transplant candidates are screened if medical factors like blood type or weight make them incompatible with an organ donor. The new allocation algorithms will also be flexible enough to account for factors unique to each organ type. For example, immune system compatibility is important when matching kidney donors to recipients

Changes made to OPTN’s decision-making framework will affect all organ donations in the country: every transplant hospital, organ procurement organization and histocompatibility lab in the U.S. is connected through a nonprofit organization that supports OPTN in partnership with the federal government. Not only does the national organ allocation system have life-or-death implications for many patients across the country, but it also has an important role in shaping systemic issues of access and equity in American healthcare. Recent research has shown how prior models have led to a disparity in the care that African-American chronic kidney disease patients receive, including transplantation access, for example. Women also had less access to kidney transplantation compared to white men under the prior model. Like other attempts to maximize efficiency of limited resources using data-driven analytics, academics warn against the ethical issues  that may arise with algorithm-based organ allocation decisions. For example, programs listing liver transplant candidates were able to game a previous algorithm used to prioritize liver donations, and a previous proposal for a kidney allocation algorithm based primarily on longevity would have violated the Age Discrimination Act. Considering ethics upfront — designing allocation models around metrics of not only efficiency but also ethics — is particularly important given the high-stakes implications of organ allocation.

Journalists can follow along with organ-specific updates from OPTN, the Department of Health and Human Services, and other organizations to cover these algorithms as they continue to be adopted through 2023. Journalists can also research the legal and regulatory history of organ distribution in the U.S., community input considered in the development of the continuous distribution model, and tools related to organ allocation. For example, OPTN made an interactive dashboard to simulate comparisons and match runs. Furthermore, while organ allocation is organized and overseen at a national level, journalists could consider how this new framework impacts states, local communities, hospitals, and individuals, such as by investigating doctors’ and patients’ criticisms of the new systems. Journalists could also consider how these changes to OPTN occur in the context of recent policies concerning organ procurement organizations included in the national network. 

School districts use machine learning to identify high school drop-out risk

On-time graduation is an important metric for public high schools across the United States. Large “graduation gaps” point to the inequities and shortcomings of the American education system, and federal law requires states and districts to report high school graduation rates and intervene in schools with low rates. Improving drop-out rates is difficult, however, as school counselors are tasked with large caseloads and identifying at-risk students requires context and time. Many state and local agencies have adopted data-driven modeling tools to address these challenges, including the Kentucky Department of Education’s Early Warning System

The Kentucky Department of Education (KDE) developed the Early Warning System, an automated, machine learning based tool, in collaboration with Infinite Campus, a software company that hosts the state’s student data entry system. Based on this continually updating source of student information, the Early Warning System uses machine learning to measure how risk factors (such as attendance, academics, and home stability) predict graduation. The system automatically scores each student’s likelihood of graduating on a scale from 50 to 150, which indicates high, medium or low drop-out risk (the lower the number, the higher the risk). The Early Warning System’s interactive interface allows educators to view, filter and search these risk assessments in real time to ensure each student receives the necessary support. A visual dashboard allows users to view overall score distributions at various levels to help district and school personnel better understand what policies yield the greatest impacts on graduation. 

Though the Early Warning System began in KDE, Infinite Campus serves over 2,000 school districts across 45 states, and it made its Early Warning System available in additional states beginning in 2019. Michigan, Montana and Sheridan County School District #2 in Wyoming are among the other state and local agencies using this system. Other government organizations have adopted similar approaches: as early as 2013, 26 jurisdictions used early warning reports to identify students at risk of dropping out. 

Like other information technology software, these tools present privacy concerns. The student data used in these assessments contain personally identifiable information that is covered by privacy laws, such as the federal Family Educational Rights and Privacy Act. Though Infinite Campus’ Early Warning System doesn’t store individual student information, confidential student data have been surreptitiously used in the past. A 2020 Tampa Bay Times investigation, for example, uncovered that a sheriff’s office used school district data to label children as potential future criminals. Furthermore, the Early Warning System learns risk factors based on population-level trends, which could actually result in biases against some demographic groups. A student’s stability rating (a subsection of the overall risk assessment) takes into account information about race/ethnicity and gender, for example. 

Additionally, even if these efforts effectively mitigate dropouts, they do not fully address general criticisms about legislation based on graduation rates. The national graduation rate has increased since 2002, reaching approximately 88% in the 2017-2018 academic year. However, critics question whether this may reflect regulatory policy incentivizing graduation over quality of education. Reports have also shown how using graduation rate as a metric overlooks schools that struggle to advance students through high school and how low-income and students of color are disproportionately affected by this. 

As education agencies throughout the country have already adopted various drop-out risk assessment systems, journalists can begin by reviewing news coverage and studies of existing algorithms, such as public evaluations of the interventions that individual agencies implement for different risk ratings. Many organizations provide resources for educators to navigate these systems, which could also be useful to journalists. Similarly, state education department websites post state-specific plans related to federal education laws and information on student privacy protections. In addition to monitoring other clients that adopt Infinite Campus’ Early Warning System, journalists can watch for new drop-out risk assessment systems. Furthermore, while Kentucky has one of the highest average high school graduation rates in the country — the state had a 4-year graduation rate of 91.1% in 2020 — journalists might consider researching early warning systems in low-performing high schools.

Algorithmic pretrial risk assessment may just be more common than you think

Californians recently voted to reject Proposition 25, which sought to replace cash bail throughout the state with algorithmic risk assessments. But, like it or not, government agencies are already moving forward with algorithmic pretrial reform efforts. For example, the Los Angeles Superior Court piloted a program in March 2020 that utilizes a tool to calculate a defendant’s risk of failing to appear in court and recidivating pretrial. Outside California, the New York City Criminal Justice Agency has a similar release assessment that draws on data from over 1.6 million previous cases to calculate a risk score that informs judges’ pretrial decisions. And communities like Pierce County in Washington State are working with the National Partnership for Pretrial Justice to develop, implement and research pretrial risk assessment systems

Proponents of pretrial risk assessment argue that algorithms can be used to address issues of mass incarceration, inefficiency, and inequity in the criminal justice system. The aforementioned pilot program in Los Angeles was used to rapidly reduce the county’s incarcerated population in response to the COVID-19 pandemic, for example. The New York City Criminal Justice Agency said its release assessment could help alleviate the city’s backlog of pending cases, according to recent Wall Street Journal coverage, and the National Partnership for Pretrial Justice similarly hopes to use risk scores to support fairness in judicial decision making

More generally, according to a 2020 report by the Brennan Center, over 70% of the American prison population (about 536,000 people) are pretrial detainees, and many of these unconvicted individuals are only detained while awaiting trial because they can’t afford bail. Making pretrial detention decisions based on data-based risk assessments rather than ability to pay bail would stop this system of wealth-based discrimination, according to proponents of California’s Proposition 25, who hoped to implement pretrial assessment systems and eliminate money bail throughout the state. 

Others, however, argue that pretrial risk assessments do not help judges make more accurate, unbiased decisions. Opponents of such systems include not only those that oppose eliminating money bail (such as the bail bond industry and some law enforcement agencies); rather, many civil rights organizations that advocate for criminal justice reform are also against the adoption of pretrial risk assessments. In 2018, a coalition of over 100 civil rights, digital justice, and community-based organizations published a statement of concerns about embedding algorithmic decision making in the criminal justice system. 

Many academics also echo this skepticism. In 2019, 27 prominent researchers signed an open statement voicing concerns over “serious technical flaws” that undermine the accuracy, validity and effectiveness of actuarial pretrial risk assessments. More specifically, like many civil rights advocates, they argued such systems cannot adequately measure the risks that judges decide on. Instead, computer-based risk evaluations ultimately perpetuate historical racial inequities in the criminal justice system. 

Some government agencies and risk assessment developers have made efforts to bring transparency to these pretrial systems, so researchers and journalists could first search for readily available information before filing Freedom of Information Act requests. Legislation that would implement more of these algorithms is also something to keep an eye on. California’s Proposition 25, for example, presented the possibility that every county in the state would have to adopt pretrial assessment systems, each of which would have been important to examine in detail. Furthermore, computer-based risk assessments are also used in other areas of the criminal justice system, including recidivism reduction algorithms used at the federal level

Government agencies, big and small, are increasingly adopting controversial algorithms for hiring

As with private companies and nonprofit organizations, government agencies—both big and small—are adopting automation in Human Resources (HR) decision-making processes. For example, federal and local government agencies use algorithms to handle leave requests, issue certifications, and run background investigations. Algorithms, however, don’t eliminate discrimination and other inequities in hiring processes; rather, they can even exacerbate or mask them

HR Avatar is one example of a system that government agencies use to incorporate automation in HR decisions. The system uses AI-driven voice and personality analysis of tests and interviews to provide a quantitative evaluation of applicants for over 200 different positions, which a range of government agencies use to compare, screen and select applicants. Federal and local agencies use HR Avatar for these pre-employment assessments: the Department of Homeland Security, Transportation Security Administration, Federal Aviation Administration and Department of Commerce are listed as clients on HR Avatar’s website.

Recently, the Pottawattamie Sheriff Office, in Iowa, listed a call for applicants to a detention officer position and explained that this system would be used in the hiring process. Applicants first participate in HR Avatar’s Correctional Officer Pre-Employment Assessment and Virtual Interview, which provides the county sheriff office with an overall score and summary of each candidate, competency scores in various areas important to the position and other evaluations. The office then uses these evaluations in the selection of applicants to invite to continue in the hiring process for this position. In effect, HR Avatar’s standardized, automated evaluation of soft skills helps the county make hiring decisions about a position that requires fluency in social skills, such as the ability to retain composure when dealing with violent or hostile individuals.

Although automation can improve the efficiency and success of hiring processes, and artificial intelligence is increasingly enticing as companies continue to shrink and outsource HR departments, researchers have highlighted the challenges in using data science for HR tasks. Automated processes have the potential to reflect existing biases, proactively shape applicant pools and otherwise perpetuate discrimination in hiring practices. While there are many employment laws that address discrimination in hiring practices, the issue of identifying and mitigating discrimination in employment screening algorithms raises new policy concerns.  To further investigate government use of HR Avatar, journalists can issue FOIA requests to agencies that use the system. While investigating HR Avatar’s assessment system may be difficult since HR Avatar is a private company, journalists could review public documents. To further investigate government use of HR algorithms in general, journalists can research state and federal laws (and proposed legislation) about automated employment practices. Furthermore, other leads in the Algorithm Tips database point to automation in HR decision-making processes, including applicant selection at a city fire department, the Department of Justice, and the U.S. Armed Forces.

With climate-related floods on the rise, FEMA is updating an algorithm that impacts 96% of flood insurance in the U.S.

The National Flood Insurance Program (NFIP) has set insurance rates in the same way since the 1970s. Over the last fifty years, however, the program has faced mounting financial issues and criticism from policymakers, fiscal conservatives, environmentalists and other stakeholders. In response, the Federal Emergency Management Agency (FEMA), the department that manages the NFIP, recently announced a new system of insurance rating: Risk Rating 2.0 is currently scheduled to be implemented in October 2021.  

According to FEMA, floods are the most common and costly natural disasters in the United States. Flood insurance is not included under standard homeowner and renter insurance, and since the market for private flood insurance is relatively small, NFIP currently provides over 96% of flood insurance in the U.S. As of December 2019, the NFIP had over 5 million policies providing over $1.3 trillion in coverage. 

However, the program has struggled to remain fiscally solvent while providing affordable flood insurance. According to the Government Accountability Office, FEMA’s debt stood at $20.5 billion in September 2018 despite Congress cancelling $16 billion in debt the year before. The Government Accountability Office has designated the NFIP as “high risk” since 2006, because emphasizing affordability created cases where premium rates did not reflect the full risk of loss and produced insufficient premiums, which in turn transferred the financial burden of individual property owners to taxpayers as a whole. Additionally, scientists have criticized this insurance rating system for reinforcing risky patterns of development.

The goal of the NFIP’s redesigned insurance rating system is to incorporate modern flood risk assessments, including private-sector data, to deliver rates that are “fairer, easier to understand, and better reflect a property’s unique flood risk,” according to FEMA. These changes also need to address FEMA’s funding issues, so although legislation limits annual premium increases to 18%, some policyholders and policymakers of heavily affected areas expressed concern over potential premium increases.

In addition to the direct impact this new rating methodology will have on the NFIP’s five million policyholders, Risk Rating 2.0 will also indirectly affect taxpayers who contribute to national programs like NFIP, and more accurate risk assessments will hopefully discourage risky development. Furthermore, the impact and importance of the NFIP will only continue to grow as global climate change continues to increase high-tide flooding in American coastal communities. 

To investigate this issue, researchers and journalists can review the Government Accountability Office’s reports on the NFIP and FEMA’s decision to postpone implementation of Risk Rating 2.0 from 2020 to 2021. Freedom of Information Act requests could be filed with FEMA to learn more about the new methodology and about what private-sector data will be used in the updated methodology. Much of the government data on flood risks are the products of other algorithms. For example, FEMA uses software like the Wave Height Analysis for Flood Insurance Studies and the Flood Risk Map to evaluate flood risk in low-lying landscapes and other areas. Investigating these tools could provide a more thorough understanding of the data and models that inform  Risk Rating 2.0.