In this section, we review best practices for communicating public health recommendations based on modeling and analytics.
Many public health practitioners communicate about public health issues on a daily basis with public health colleagues, clinicians, elected officials, and the public.
When communicating about modeling results, be mindful of all of your standard public health communication training, as these approaches are still relevant.
Keep the focus on the public health issue that is affecting individuals and communities -  Not the details of modeling methods. 
Distill your message into a single overriding communication objective, or SOCO, and state it clearly and often.
As with any public health recommendation, be transparent about your level of confidence, and when more information might be available.
Communicate the specific nuances of modeling results. By definition, modeling outputs - unlike public health surveillance data - are not simply observations, and are often more uncertain. Communicating this uncertainty requires particular skills.
If sharing specific modeling results, explain why the model was needed.  Clarify whether the model supports planning, explores outcomes of potential actions, or addresses data gaps. Explain that we don't know the future and that models rely on past patterns; if those patterns change, uncertainty increases.
Because of this uncertainty, avoid overly specific results that imply precision, like: "there will be 16,473 hospital admissions for respiratory viruses" even if that is what your model output said. Instead, focus on the topline messages, such as: "we expect hospital demand to exceed bed capacity".
Most of infectious disease models rely on real world data, which are imperfect. The public understands this, and appropriately acknowledging limitations in the underlying data helps build trust in the modeling output. 
In addition to potential issues with the underlying data, models themselves make assumptions that impact results. Being open about major modeling assumptions is critical to building trust in the output and providing a frame of reference for updates. For example: "we assumed no new virus variant would emerge" or "we assumed flu vaccine efficacy would be similar to last year". That enables you to identify what might change your assessment. Some assumptions can be updated as more data become available; like if a virus variant did emerge, you may wish to update your model and change your assessment.
The communication approaches discussed earlier apply across audiences. Elected officials, executive decision-makers, healthcare providers, and the public all want  a clear focus on the public health issue, a single overriding message, and an understanding of data limitations and assumptions. Your audience will help you shape your modeling question and how you communicate results. Meaning, consider your audience at all phases, not just after you have results. Technical detail may vary, but always: maintain a public health focus; present a clear SOCO; and be transparent about assumptions and limitations.
Here are some tips to communicate about model outputs. Emphasize the overall public health message, instead of modeling methods. For example, "We are expecting a severe respiratory viral season...". Acknowledge uncertainty with language like, "This assessment is based on what we know now and may change as we learn more ..." Use plain language rather than jargon by saying, "To generate these results, we studied past data and current trends ..." And, engage your partners in coordinating messaging. Communicate this by saying, "We are working closely with healthcare providers across the state to prepare ...". 
Members of the media can be useful partners for communicating with the public. In some ways, the media is just another audience, so remember your general communication principles. Have a clear message, and stay focused. Be clear about what is important for the public to know or act upon by homing in on the 'so what?' And as always, do not use jargon.
Remember these principles during the next part of this training series where we will apply them to an example.