SubHero Banner
Social Share
Article In-page Navigation
Container for Section 1

Imagine if epidemiological forecasting could predict serious outbreaks of disease, just like we predict hurricanes, snowstorms and tornados. It could guide decisions of government and health care officials, business leaders and the general public. And just like an accurate weather forecast, it could help protect the health and well-being of an enormous number of people. That’s the potential for the Optum flu forecast, a data-driven, early warning system for influenza and influenza-related illnesses.


The economic and health burdens of influenza

Influenza, more commonly known as the flu, has a tremendous impact on the United States each year. Estimates place the annual economic burden of this contagious respiratory virus at $11.2 billion.1 This includes $3.2 billion in direct medical costs and $8 billion in indirect costs, such as lost business productivity. The human burden is even more significant. The flu affects millions of people each year, with many becoming seriously ill. The Centers for Disease Control and Prevention (CDC) estimates that in the 2017–2018 flu season, an especially bad one, the virus caused roughly 49 million illnesses, 960,000 hospitalizations and 79,000 deaths.2

Vaccination and other preventive steps reduce the risk of contracting and spreading the flu but don’t entirely eliminate it.3 Data and analytics advancements, however, offer another tool to further prevention efforts — predicting when and where increased flu activity will occur. With these insights, we can support efforts to mobilize a response and lessen the flu’s health and economic impacts.


Accordion Block v2
  • 2017-2018 Flu season burden2

    Human Impact

    • 79 thousand deaths
    • 960 thousand hospitalized
    • 49 million illnesses


    Economic impact

    • $3.2 billion direct costs
    • $8 billion indirect costs



Container for Section 2

Academic research and added Optum health care intelligence lead to an innovative health care solution

The flu forecast model grew out of a relationship between Optum and Carnegie Mellon University (CMU), an academic institution renowned for its computer science and artificial intelligence (AI) research.

Today, corporate support has become even more critical for university researchers. “We’ve seen increasing interest in research collaborations between academia and industry due to tighter government funding of academic research and the desire for companies to learn from and work with the best scientific and technology university talent in areas that align with enterprise priorities,” says Paul Nielsen, vice president of Strategic Programs in the Optum Advanced Technology Collaborative, part of UnitedHealth Group, where he leads a group focused on strategic partnerships with industry and academic research institutions.

Optum has provided a gift to support the research of the Delphi group at CMU, headed by Roni Rosenfeld, PhD, an expert in epidemiological forecasting and head of the university’s Machine Learning Department. Professor Rosenfeld has also been advising Optum on its development of an early warning system for influenza and influenza-related illnesses, which is based on CMU-developed technology made available via a General Public License.


This is an example of how the health care industry and research institutions can employ advanced technology to impact the health of millions of people. We can alert clinics in, say, Dallas that the flu is coming three weeks from now. They can then stock up on Tamiflu and start lining up extra staff.

– Paul Nielsen, Vice President of Strategic Programs in the Optum Advanced Technology Collaborative


By bringing that solution inside Optum, its forecasts can make an impact throughout the larger UnitedHealth Group universe, from alerting at-risk populations to an incoming flu to helping guide internal operations. Teams at Optum applied their expertise and curated data assets to expand the technology’s scope and capabilities. With this added health care intelligence — something Optum calls OptumIQ™ — the Optum flu forecast was born.

“This is an example of how the health care industry and research institutions can employ advanced technology to impact the health of millions of people. We can alert clinics in, say, Dallas that the flu is coming three weeks from now,” says Paul Nielsen, vice president of Strategic Partnerships in the Optum Advanced Technology Collaborative, where he leads a group focused on strategic partnerships with industry and academic research institutions. “They can then stock up on Tamiflu and start lining up extra staff.”


Container for Section 3

Data and technology lay the foundation for accurate predictions

The Optum flu forecast relies on an extensive array of data to predict flu outbreaks down to a metro and county level two to four weeks before they occur. This data includes de-identified medical and pharmacy claims records related to flu diagnosis from Optum, along with social data — where people are born, grow, live, work and age — digital surveillance data and clinical reports publicly available through the CDC. Additional information from non-Optum sources, such as flu-related web search queries and surveillance information available publicly add further insight and increase forecasting accuracy.

“Any one of these data sources on their own would not necessarily be that great, but all of them together are quite good and quite robust,” notes Professor Rosenfeld. The data from Optum is some of the best he has worked with, he says. And unlike some other data sources, it’s geographically specific, not aggregated at a state level. Accurate and extensive, its “noise” level is relatively low.

Since 2014, Carnegie Mellon University’s flu forecasting technology and systems have won an annual flu forecasting competition headed by the CDC. Extending the university’s forecast work, Optum is confident it can predict with a high level of probability where flu hot spots will be in specific metro areas, supporting business decisions for the upcoming flu season.


Container for Section 4

The benefits of flu forecasting

In the past, tracking existing flu activity has guided efforts to combat the virus. However, that tracking usually lags behind what’s happening in real- time. Accurately predicting when and where flu activity will increase is a game-changer. It can enable us to better guide prevention efforts and health system readiness to protect patients, members, employees and communities at large — saving lives and dollars. Specifically:


Health officials can best mobilize resources

Vaccines, diagnostic test kits and personal protective equipment are essential tools for combating a severe flu outbreak. The flu forecast model will help UnitedHealth Group (UHG) in its efforts to make sure there are enough of these and other supplies on hand and to move them to areas of greatest need.




People in at-risk areas can be encouraged to get vaccinated and take steps to reduce flu transmission

Vaccination is the best-known way to reduce the risk of contracting flu, but it takes about two weeks for the vaccine to offer protection against the virus.4 Currently, the CDC recommends nearly everyone 6 months of age and older be vaccinated by the end of October each year.5 Vaccination later in the season still lowers the risk of contracting the flu or becoming seriously ill from it, but many people never do get vaccinated. In the 2017–2018 season, only 37.1% of adults and 57.9% of children had been vaccinated by season’s end.6

During the 2017–2018 flu season, the CDC estimates that flu vaccination prevented 7.1 million illnesses, 109,000 hospitalizations and 8,000 deaths.7 Unfortunately, the 2017–2018 season also brought the highest number of deaths in over a decade and the most deaths for children since the CDC began tracking pediatric deaths — more than 180.8


Accordion Block v2
  • Benefits of flu vaccination during the 2017-–2018 influenza season7

    • 7 million: The estimated number of flu illnesses prevented by vaccination (about the population of New York)
    • 109,000: The estimated number of flu hospitalizations prevented by vaccination (about the number of vehicles crossing the Golden Gate Bridge each day)
    • 8,000: The estimated number of flu deaths prevented by vaccinations (Twice the number of hospitals in the U.S.)


With more than two weeks’ advance notice of an outbreak in a specific area, people can be reminded to get a flu vaccine, even late in the season, which is often when there is another peak in flu activity. Optum flu forecast predictions will provide insights to better target and time communications to individuals before the onset of flu in a local area. The timing is important. Flu prevention strategies, including vaccination, are commonly communicated to the public, but because the flu season may last half a year or longer, those messages can become commonplace and easy to disregard, resulting in “flu prevention fatigue.”

When the location and time frame of a specific flu threat can be communicated, the public may be more likely to take heed. Knowing when and where to target public education efforts may reduce the flu information fatigue later in the season and be more effective in encouraging people to take additional steps to lessen flu activity. Messages can communicate preventative measures, such as washing hands frequently, avoiding crowded public areas and people who are sick and staying home if symptomatic.9 These messages can also be customized for the specific population segment.


Providers, clinics and hospitals can prepare for an influx of flu-related illness

Anyone can experience flu-related complications, but those most at risk include young children, pregnant women, people 65 years of age and older, and those with certain chronic health conditions.10 Forecasting allows health officials to focus limited resources on areas where it counts, ensuring adequate supplies of antiviral medications, respiratory hygiene items, health care workers and hospital beds for those who get sick. Health care facilities may also take precautions such as separating people with flu symptoms from others in waiting rooms or providing face masks to minimize the risk of transmission.11

Antiviral drugs may lessen flu symptoms, shorten its duration and reduce the risk of flu-related complications, but they work best if given within two days of the onset of symptoms.12 The flu forecast model can also support distribution strategies to get adequate supplies of these medications to regions of the country where they are needed, preventing shortages as well as overstocks that lead to shelves of expired medicines.

Additionally, retailers can use flu forecasts to ensure supplies of over-the-counter medications such as pain relievers and anti-fever medicines, as well as products like thermometers, will be available to their customers.


Employers can take steps to ensure business continuity

Whether in the health care industry or another business setting, employers can be better prepared for a flu outbreak . Appropriate precautions may prevent them from losing a large number of employees to illness and enable them to continue normal operations. The flu usually spreads when someone comes in contact with tiny droplets of moisture that an infected person expels when coughing, sneezing or speaking.13 People can also contract the virus by touching it on a surface and then touching their hand to their nose or mouth.14 A variety of precautions may lower transmission risk.

At UHG, for example, precautions might include conducting virtual, rather than in-person, meetings to reduce the opportunity for flu transmission at the workplace or adjusting work arrangements to reduce the density of employees in confined spaces, like call centers. UHG also takes precautionary measures, including increasing messaging to employees about proper hand hygiene and cough etiquette, initiating work-from-home protocols, offering on-site vaccination clinics, discouraging employee travel to outbreak areas and increasing workplace sanitization efforts. The care delivery organizations within Optum may also opt to conduct telephonic rather than face-to-face patient assessments, when possible, to better protect both the patient and the employee.

Many of these actions and policies are too onerous to sustain for the entire national flu season, which lasts many months. However, they could be used cost-effectively in particular cities during the few weeks around the peak of a local epidemic.


Individuals can choose to act on flu warnings

Just as people may check the weather forecast to decide if they should take an umbrella when leaving home, individuals, especially those most at risk for flu complications, may check the flu forecast to decide whether to leave their home at all.


Container for Section 5

Looking ahead

The Optum flu forecast illustrates how advances in data curation and emerging technologies can uncover new health care intelligence that industry stakeholders can use to lower costs, create better health outcomes and improve consumer experiences. Predictive insights won’t eliminate the flu, but they can change our ability to prevent and combat it. And it may be just the beginning of our capacity to influence the trajectory of infectious diseases and their impact on the U.S. population. With surges of other vaccine-preventable illnesses, such as measles, pertussis and varicella regularly occurring throughout the country, the potential uses for epidemiological forecasting are significant.

Optum stands ready to take the next step. Driven to a commitment to create actionable insights, we strive to solve the biggest problems in health care and work across all parts of the system to make it work better for everyone.



It's amazing to see what data, analytics and expertise can do. Explore how infusing OptumIQ into all our products and services powers intelligence across the health care system.

Explore OptumIQ



  1. Putri WCWS, Muscatello DJ, Stockwell MS, Newall AT. Economic burden of seasonal influenza in the United States. Vaccine. 2018;36(27):3960-3966. doi:10.1016/j.vaccine.2018.05.057. Accessed June 25, 2019.
  2. Centers for Disease Control and Prevention. Disease burden of influenza. Last reviewed February 19, 2019. Accessed. June 30, 2019.
  3. Centers for Disease Control and Prevention. Preventive steps. Last reviewed November 13, 2018. Accessed June 29, 2019.
  4. Centers for Disease Control and Prevention. Key facts about seasonal flu vaccine. Last reviewed September 6, 2018. Accessed June 26, 2019.
  5. Ibid.
  6. Centers for Disease Control and Prevention. Early-season flu vaccination coverage — United States, November 2018. Last reviewed December 14, 2018. Accessed June 26, 2019.
  7. Rolfes MA, Flannery B, Chung JR, et al. Effects of influenza vaccination in the United States during the 2017-–2018 influenza season. Clin Infect Dis. Published February 2, 2019. Accessed June 30, 2019.
  8. Center for Disease Control and Prevention. Summary of the 2017-–2018 Influenza Season. Last reviewed November 2, 2018. Accessed July 25, 2019.
  9. Centers for Disease Control and Prevention. Preventive steps. Last reviewed November 13, 2018. Accessed June 29, 2019.
  10. Centers for Disease Control and Prevention. Key facts about influenza (flu). Last reviewed August 27, 2018. Accessed June 22, 2019.
  11. Centers for Disease Control and Prevention. Prevention strategies for seasonal Influenza in healthcare settings. Last reviewed October 30, 2018. Accessed June 29, 2019.
  12. Centers for Disease Control and Prevention. What you should know about flu antiviral drugs. Last reviewed April 22, 2019. Accessed June 26, 2019.
  13. Centers for Disease Control and Prevention. Key facts about influenza (flu). Last reviewed August 27, 2018. Accessed June 22, 2019.
  14. Ibid. 

Tag: Articles and blogs, Providers, Articles, AE 20.7 OptumIQ, Articles, data and analytics capabilities, Health plans, OptumIQ

Cross Promo Article(S)
Resource Library Side Bar

Get the latest in health care intelligence delivered to your inbox.