Can Lyme Disease Risk Be Predicted Years in Advance?
Tick populations reflect conditions over several years
Weather and wildlife data may provide an early warning
Regional forecasts could strengthen Lyme disease prevention
Most people think about Lyme disease risk in terms of the current season. Was the winter unusually warm? Has the spring been rainy? Are ticks especially active this summer?
Those questions matter, but they may capture only part of the picture. Blacklegged ticks develop through a life cycle that generally lasts about two years. The tick encountered today may therefore reflect environmental and wildlife conditions from an earlier year.
A 2026 study from Minnesota examined whether climate, snow, soil moisture, drought, temperature, and small-mammal populations could help predict future tick abundance and human Lyme disease cases. The researchers reported that environmental data predicted high tick abundance most accurately approximately one year later. A machine-learning model performed best when predicting relatively high Lyme disease case years using conditions from two years earlier.
The findings raise an important public-health question: could communities receive an early warning about an elevated Lyme disease season before patients begin arriving with tick bites, rashes, facial palsy, joint swelling, or other manifestations of the infection?
Why Might Lyme Disease Risk Be Predictable?
Lyme disease risk does not depend on a single variable. It develops through interactions among ticks, animal hosts, the Lyme bacterium, local habitat, weather, and human behavior.
Blacklegged ticks pass through egg, larval, nymphal, and adult stages. After hatching, they must obtain a blood meal at each stage to survive. Temperature and moisture can affect their development and survival, while snow may insulate ticks and the ground-level habitat from severe winter conditions.
Animal populations matter as well. White-footed mice are highly competent reservoirs for Borrelia burgdorferi, the principal bacterium that causes Lyme disease in the United States. A larval tick that feeds on an infected animal may acquire the bacterium and later transmit it as a nymph.
This sequence creates a delay. Environmental conditions that influence mice or immature ticks in one year may affect the number of infected nymphs—and potentially human disease—in a later year.
What Did the Minnesota Study Examine?
The researchers combined several types of information from seven counties in the Twin Cities metropolitan region:
- Active tick surveillance from 2014 through 2023
- Small-mammal trapping data
- Reported Minnesota Lyme disease cases
- Precipitation and soil moisture
- Snow depth and snow water equivalent
- Drought conditions
- Temperature and accumulated degree days
- The number of extremely cold days
- The proportion of white-footed mice among trapped small mammals
During the study period, surveillance identified 18,071 ticks. Approximately 76% were Ixodes scapularis, the blacklegged tick associated with Lyme disease in the Upper Midwest. Most of the collected ticks were larvae, with the remainder being nymphs. No adults were included.
The investigators then asked whether these environmental and host variables were associated with relatively high or low tick years and relatively high or low Lyme disease case years. They evaluated conditions in the same year and after one- and two-year delays.
They used both conventional statistical models and gradient-boosting machine learning. The machine-learning approach can examine complex relationships among many variables, but its results still require careful validation and interpretation.
What Predicted a High-Tick Year?
No single weather measurement explained tick abundance. Moisture, precipitation, snow, temperature, drought conditions, and small-mammal measures contributed in different ways depending on the model and the length of the time lag.
Across the analyses, signals repeatedly appeared for drought severity, soil moisture, snow, extremely cold days, degree days, the number of small mammals, and the proportion of white-footed mice.
The models predicting tick abundance generally performed better when environmental conditions from the previous year were considered. This supports the biological expectation that a tick population developing through several life stages will not respond only to the weather occurring during the week or month when people notice ticks.
The direction of some associations was not always intuitive or consistent. For example, moisture or snow measures did not have the same relationship in every model. That is one reason the study should not be reduced to a simple rule such as “more snow always means more ticks” or “drought always means fewer Lyme cases.”
Could Lyme Disease Cases Be Predicted Two Years Ahead?
The study’s most striking result involved human Lyme disease cases. In the machine-learning analysis, the ability to distinguish a relatively high-case year improved as the time lag increased and was highest with a two-year lag. The reported area under the receiver operating characteristic curve was 0.92.
An area under the curve, or AUC, measures how well a model distinguishes between two categories. In this study, those categories were high- and low-case years based on whether the annual case count was above or below the median. An AUC of 0.92 indicates strong discrimination within the dataset that was analyzed.
However, it does not mean that the model predicted the exact number of Lyme disease cases with 92% accuracy. It also does not establish that a two-year forecast would perform equally well in a future Minnesota season or in another state.
The conventional backward-selection models did not identify delayed predictors of Lyme case years in the same way. The strongest delayed result emerged from models built using the machine-learning approach. This difference makes prospective testing especially important before the method is treated as an operational forecast.
Why Were White-Footed Mice Important?
The researchers considered both the total number of trapped small mammals and the proportion that were white-footed mice. The proportion may matter because an ecosystem dominated by a highly competent reservoir host can offer larval ticks more opportunities to acquire B. burgdorferi.
By contrast, a more diverse group of animal hosts may include species that are less efficient at passing the bacterium to feeding ticks. This idea is sometimes described as a dilution effect, although the strength and operation of that effect can vary among ecosystems.
Mouse abundance alone is therefore not a complete measure of Lyme disease risk. The composition of the local animal community, tick survival, infection prevalence, vegetation, and human contact with tick habitat all influence whether an ecological signal eventually becomes a human case.
How Could an Early-Warning System Help?
If these findings are validated, state and local health departments might use environmental and wildlife surveillance to prepare before a high-risk season rather than responding only after cases increase.
An early warning could support:
- Earlier and more intensive public education about tick avoidance and removal
- Targeted prevention messages in communities expected to have higher exposure
- Additional tick surveillance in locations where the model signals increasing risk
- Reminders to clinicians to consider Lyme disease when compatible symptoms develop
- Planning for diagnostic, laboratory, and public-health resources
- Evaluation of prevention measures before the predicted high-risk period
This approach reflects a One Health model: human disease is considered together with animal hosts, tick populations, and environmental conditions.
Effective forecasting would not eliminate the need for real-time surveillance. A multiyear signal could identify a period requiring closer attention, while current tick counts, pathogen testing, patient reports, and case surveillance could show whether the anticipated risk is actually emerging.
What Are the Study’s Important Limitations?
This was a valuable regional analysis, but several limitations prevent broad conclusions.
The data came from seven counties in metropolitan Minnesota. Tick biology, seasonal activity, wildlife populations, climate relationships, and human behavior can differ substantially between the Upper Midwest, Northeast, South, and West. A model developed in Minnesota should not be assumed to work in New York, North Carolina, or California without local validation.
The tick-surveillance program collected ticks from trapped small mammals and primarily captured larvae. It did not directly measure all host-seeking ticks that might bite people. Human cases were assigned according to county of residence, although infection may have occurred during travel or in another county.
The Lyme disease surveillance definition also changed in 2022, and case surveillance does not identify every infection. The study lacked human case data for 2020.
In addition, the analysis included a relatively limited number of counties and years for training models with many possible predictors. Machine-learning methods can detect useful patterns, but they can also fit relationships that do not perform as well when applied to new data. Prospective validation is needed to determine whether the model can predict a high-risk year before it occurs.
What Does This Mean for Patients Now?
The study should not change an individual patient’s immediate prevention decisions. A forecast cannot determine whether a particular yard, trail, trip, or tick carries B. burgdorferi. Even during a relatively low-tick year, a single infected tick can transmit disease.
People should continue using practical precautions in areas where ticks may be present. These include avoiding direct contact with brush and leaf litter when possible, using appropriate repellents, treating clothing or gear as recommended, checking the body after outdoor exposure, showering after returning indoors, and removing attached ticks promptly.
A person who develops an expanding rash, facial weakness, flu-like illness, new joint swelling, or neurologic or cardiac symptoms after possible tick exposure should seek medical evaluation. The absence of a public warning for a “bad tick year” should not be used to dismiss a compatible clinical presentation.
The larger implication is for public health. Environmental forecasting may eventually help communities act sooner, but it should complement—not replace—clinical assessment, current surveillance, and individual prevention.
Frequently Asked Questions
Can weather predict how bad a Lyme disease season will be?
Weather and environmental conditions may contribute to a forecast, but no single measurement can reliably define the season. Tick abundance and human cases also depend on wildlife hosts, habitat, tick infection rates, human behavior, and conditions from earlier years.
Why might conditions from two years ago affect Lyme disease today?
Blacklegged ticks generally require about two years to progress through their life cycle. Conditions affecting animal hosts, larval feeding, development, and survival can therefore influence the infected nymphs people encounter in a later year.
Did the Minnesota model predict the exact number of Lyme disease cases?
No. The study evaluated whether a year had a relatively high or low case count based on the median. Its reported AUC measured the model’s ability to distinguish those two groups, not the percentage of individual cases predicted correctly.
Can this model predict my personal risk of getting Lyme disease?
No. It was a county-level ecological model and cannot determine whether an individual will encounter an infected tick. Personal risk depends on location, activity, prevention practices, tick attachment, and other factors.
Can the Minnesota findings be applied throughout the United States?
Not yet. The relationships need to be tested prospectively and in other geographic regions. Local climate, habitats, tick populations, animal hosts, and patterns of human exposure may produce different results.
Clinical Takeaway
A Minnesota study found that climate, snow, soil moisture, temperature, and small-mammal data may help identify high-tick and high-Lyme disease years before they occur. Tick abundance was predicted most accurately with approximately a one-year lag, while a machine-learning model distinguished high Lyme case years most accurately using a two-year lag.
The model remains regional and requires prospective validation. It cannot predict an individual infection or replace real-time tick surveillance, clinical judgment, or routine prevention.
Lyme disease risk may begin developing years before a patient encounters a tick, giving public-health officials a possible opportunity to warn communities before cases rise.
Related Articles
What are the bad signs after a tick bite?
Why travel history matters in Lyme disease
Is Lyme disease moving farther south?
This article is for informational purposes only and is not a substitute for professional medical advice, diagnosis, or treatment.
References
- Angell, K. E., Jarnefeld, J., Schiffman, E. K., Broadhurst, M. J., Dong, J. J., Degarege, A., Cortinas, R., & Brett-Major, D. M. Climate and other environmental factors predict tick abundance and Lyme cases in Minnesota one and two years in advance. One Health. 2026;23:101507.
- Centers for Disease Control and Prevention. Tick lifecycles. CDC. Updated October 11, 2024.
- Eisen, R. J., & Paddock, C. D. Tick and tickborne pathogen surveillance as a public health tool in the United States. Journal of Medical Entomology. 2021;58(4):1490–1502.
- Eisen, L. Stemming the rising tide of human-biting ticks and tickborne diseases, United States. Emerging Infectious Diseases. 2020;26(4):641–647.
Dr. Daniel Cameron, MD, MPH
Lyme disease clinician with over 30 years of experience and past president of ILADS.
Symptoms • Testing • Coinfections • Recovery • Pediatric • Prevention
