Transportation and healthcare: How government agencies are leveraging AI to improve critical public service delivery

Healthcare and transportation are two of the most critical areas of public sector service delivery. Both are vital not just to individual well-being, but also underpin broader economic development, equity and resilience for the nation. As government agencies invest in insights and automation, both critical to maintain continuity of services and efficient operations in serving the public, AI is already transforming the way government and industry are meeting the mission. 

At the recent Google Public Sector Gen AI Live & Labs event, industry, education, and government experts honed in on the critical need for security, efficiency and scale that AI is delivering, as well as how Google experts and tools are working alongside the public sector to make AI a mission-revolutionizing reality — and a success.

Transforming transportation efficiency, operations, and safety

In cities across the country, Google AI for Public Sector is already at work to improve transportation, not just on roadways, but by bolstering the tools available to agents and officers in charge of everything from public safety to maintenance and operations to the reliability of public transit. 

A modern example impacting public safety and efficient operations on a large scale is on the New York City subway system, which sees 3.6 million riders per day. The Metropolitan Transportation Authority (MTA) recently worked with Google Public Sector to use sound and vibration data to test a tool called TrackInspect to proactively detect potential track defects before they escalate into service-disrupting operational issues. 

The program taps Google Pixel Smartphones fitted to R46 subway cars that detect sound and vibration data; that data is then sent through cloud-based systems in real-time and uses AI and ML algorithms to generate predictive maintenance insights based on that information. Finally, an NYC track inspector follows up with the information provided by the system to confirm whether there is an issue and provide feedback to continuously train the model.

“Just imagine you're a track inspector, which is our boots-on-the-ground inspectors. You're walking two to three miles a night in a dimly lit tunnel on an elevated structure where you're being careful that you're not stepping in any holes or anything like that, watching your footing you're inspecting with a flashlight and tools. And you also have train service on you, 24/7, 365, so the conditions that these employees work in are extremely difficult,” explains MTA Assistant Chief Track Officer Rob Sarno, noting that inspectors are also currently required to carry around a 292-page inspection manual. All-in-all, it’s a very demanding job.

With TrackInspect and the accompanying smartphone app, inspectors can streamline their efforts to areas that likely have maintenance needs, as well as carry and search through their manuals virtually. They can also tap Google’s Generative AI model Gemini to both report issues they see and search out other potential issues — and their possible causes — in the area, based on sensor data.

“It will also give them suggestions based off of our manual and our standards,” said Sarno. “It's revolutionizing … what [a track inspector] could do.”

Meanwhile, in Philadelphia, the city’s parking authority is also tapping AI with the aim to improve bus routes and times. The city was encountering issues with people parking in bus lanes and other restricted areas, like medians, which was slowing bus routes and making traffic more challenging in the city overall.

To help remedy the issue, the Philadelphia Parking Authority and the Transportation Research Board joined together to set up an AI-backed camera-assisted system to detect and automatically ticket offenders, with the hope of clearing bus lanes. After a 10-week pilot, the city recorded 36,000 violations on 10 bus routes. The city is now poised to kick off a broader pilot program on two of its slowest bus routes, said Leslie Richards, professor at Stuart Weitzman School of Design at the University of Pennsylvania and chair of the Transportation Research Board Executive Committee. 

“We know that we'll be able to increase reliability. We'll be able to get the buses faster,” said Richards. 

The city is also tapping into AI for more projects, including one that aims to help gather the data the city maintenance teams need to better clean public transit. 

Reshaping healthcare

In New York, institutions are also leveraging AI to meet the moment in healthcare, expanding what providers are able to do when it comes to delivering accurate, efficient care and improving the user experience for clinical staff and patients alike. 

At NYC Health and Hospitals, Chief Medical Information Officer Michael Bouton notes that AI is at work in many ways, including improving clinical workflows. An example is its deterioration index, which flags if admitted patients need to be transferred to the ICU or receive more intensive care.

“AI doesn't get tired, it doesn't step out to call its spouse,” said Bouton, noting that predictive models can help to attract attention where it’s most needed. “[AI] is just always on. It's always working. And that's a huge advantage.”

Meanwhile, on the non-clinical side of the house, the organization has incorporated an AI agent into its enterprise service desk experience for clinical staff, allowing them to bypass the previous system that required them to call a human agent and instead tapping an AI agent who can direct them to the correct information.

“Nobody really wants to make phone calls anymore, so I think it’s a win for everybody across our system,” said Bouton.

While it’s still early days, many institutions are starting to lay the groundwork for AI tools by building the infrastructure necessary to tap into the mounds of data healthcare has available to develop transformative AI models.

This is exactly what the experts at the Center for Population Health Data Science have been working on the last several years.

“On the heels of COVID, there was a realization that a lot of data still sits in these programmatic data stores for day-to-day program purposes. But when it comes time to take care of the full city, you need all of that data to link to one another and create a full picture for the city. We just didn’t have that,” said Mamat Parakh, deputy commissioner for the Center for Population Health and Data Science in the NYC Department of Mental Health & Hygiene.

So, the organization set out to develop the infrastructure necessary to bring data from those programmatic stores into enterprise data platforms. 

“It was a lot of freeing of the data, unlocking of the data and making it both secure and accessible and usable,” said Parakh. And as Generative AI came onto the scene, the organization also looked to the technology to bring in even more data. “We’re also in parallel utilizing AI that’s available to us to further unlock the data that’s inside our four walls but is locked in images or text. There's a lot of that in healthcare. So we have our researchers figuring out how to use language and vision models to get clean usable data out of these other data formats.”

Another example is NYU Langone Health, which has invested in AI with the aim to better predict everything from patient outcomes to operational needs. The organization recently built its own large language model trained on a decade of the institution’s clinical notes and inpatient health records with the aim to create an AI model that could utilize the organization’s unstructured data to perform tasks like categorization and numerical prediction. Called NYUTron, the model works to change “how we practice with medicine and knowledge of data,” said Nader Mherabi, executive vice president and chief digital and information officer at NYU Langone Health.

The organization is looking at integrating NYUTron into workflows across the organization, from optimizing call centers to creating discharge plans and even advising clinicians. 

And many of these use cases are just scratching the surface of what’s possible, with experts noting that the technology could be integrated into everything from cleaning and improving insurance claims with the aim to reduce denials, to offering an extra set of “eyes” to doctors looking over clinical imaging.

“It’s truly a technology that's driving disruption,” said Mherabi.

Setting a path for change

As organizations look to implement AI, however, they need more than just technology, they need adaptable tools and in-depth expertise to help ensure ethical and effective use. This is where vendor partners like Google Public Sector are already at work, collaborating across the entire spectrum of government to help guide agencies as they lay the cultural and technological foundations for AI tools. Education is key to success, ensuring users understand how AI tools work, or when they should and shouldn’t be used. Humans are still an essential part of decision-making. 

“In many ways, AI is this continuation of Google's 26-year mission to make the world's information universally accessible and useful,” said Chris Hein, field CTO at Google Public Sector during a panel on workforce. “[Education] is important and it's critical, and we do need to be making sure that we're not letting people get away with just accepting whatever the AI system that they used spat out at them. But I do think that we should assume that that's going to happen and then help them to figure out, how can they synthesize, how can they use all the knowledge that we as humans get to have access to get us to that next step of the knowledge endeavor.”

This is where partners like Google Public Sector, who have the expertise and experience to help agencies sort through the noise around AI and develop a firm foundation for AI tools that can scale effectively, can step in. And with partners that are not only knowledgeable, but also motivated to help organizations grow and change, it can transform the very way people think about government in the future.

“The good thing is, people are extremely inspired by this work … and when we realize what we can do when we partner with Google, when we partner with a startup, when we partner with other stakeholders, it's really not only fun and and helps us in our business case model, but it's what we have to do,” said Richards. “It's the only way we're going to be able to repeat it into the future and be able to relay more to our customers.”

This is the third in a series of articles based on the Google Public Sector Gen AI Live & Labs event in New York City, hosted by Google Public Sector in collaboration with GovExec. At the event, industry experts and leaders across city and state governments, as well as higher education, came together to explore how government is leveraging AI to bolster these industries, discuss use cases, and outline best practices for implementation going forward. 

Check out the other articles in this series here and stay up-to-date on the information you need to power the mission by subscribing to the Google Public Sector Newsletter.