*In the run up to Forecasting Europe  in Berlin in June, eyeforpharma asked some of our key speakers for their ins*  http://www.eyeforpharma.com/forecastingeu/
In the run up to Forecasting Europe in Berlin in June, eyeforpharma asked some of our key speakers for their insights on the state of forecasting. Heres what they had to say...
What are the key challenges facing forecasters now?
Mete Sylan, health economics and outcome strategy manager, BMS: Forecasts need to be able to link to both opportunities and threats in the market. Limiting factors in terms of delay during negotiation of prices and how changes to the structure of healthcare systems, such as the introduction of family practitioners in Turkey, need to be built into forecasts as well as opportunities such as improvements in diagnosis. (For more on Turkey, see Transforming the Turkish pharma market.)
Lauren Franzel, AVI vaccine market analyst, PATH: Forecasts need to reflect a variety of factors that impact on how vaccines can really be introduced into different countries. The concern is how cutting edge technologies in the vaccine market can be used in these countries and, at the forecasting level, to take into account local infrastructure. This means things like cold chain infrastructure, and we need to include these in our forecasts as limiting factors. (For more on vaccines in Latin America, see Mapping the Latin American vaccine market; for more on PATH and vaccines, see Why forecasts should drive strategy.)
Nic Talbot Watt, managing director, Black Swan Analysis: The market has changed over last five years. Traditionally, pharma built models based on the prevalent patient population. This essentially meant divvying up that patient population across different therapies. But with new targets, such as specialist therapy areas being developed now, we need to explore more dynamic approaches of forecasting. (For more from Nic Talbot-Watt, see Forecasting: How to get patient flow analysis right and New challenges and techniques in forecasting.)
Paul Goodwin, Department of Management Science, University of Bath: There is scope for forecasters to explore how to make the best use of judgments to improve, rather than worsen, their forecasts. That is about making the best use of forecasters and managers time to focus on the big issues that affect the forecast, not on trivial changes.
Alec Finney, principal, Rivershill: There is much more uncertainty across the globe, and so forecasters need to work harder to understand the impact of uncertainty on forecasts. They need to identify the key assumptions used in forecasting and uncertainties and explain these to colleagues. Forecasters need to identify the scenarios that could radically change the strategy for launching a product. (For more from Alec Finney, see Pharma forecasting: To offshore or not to offshore? and Unlocking the hidden value of forecasting.)
How can forecasters overcome the lack of epidemiological data?
Mete Sylan: Countries will differ in the amount of epidemiological data that is available. In some countries, like Turkey, there are efforts to collate more data, such as activity in hospitals that may also be linked to product tracking systems. But they need further development. In the meantime, we can work with market research companies and physicians to try to plug gaps in the data that forecasters need.
Nic Talbot Watt: Weve been able to provide robust forecasts even when there has been a lack of epidemiological data. We can do that by looking at surrogates from other markets and model the disease. (For more on forecasting and epidemiological data, see New challenges and techniques in forecasting and Forecasting for complex diseases.)
How should forecasters decide what models to use and how much to apply judgment?
Lauren Franzel: We work with our partners to understand the latest views on the evidence to support the use of vaccines and the market dynamics, including both the demand and supply issues. We use both to inform our forecasts. We have a sub-team that helps inform our adjustments to forecasts to account for the latest changes in policies. Our methodology and assumptions are approved and endorsed by all members of the sub-team.
Nic Talbot Watt: There is no golden rule about what model to use. Maybe its best to start with a static model but, if that is not giving you enough predictive power, move to a more dynamic model. Key is that people need to get used to the models, especially because dynamic models need careful calibration to avoid the ghost in the machine.
Paul Goodwin: Its a careful balance. In most instances, the statistical packages used by firms can provide very good and reasonably accurate forecasts. These can be adjusted from management judgment, but these changes need to be focused on big changes that can improve forecasts and avoid trivial changes that can actually worsen forecasts.
Alec Finney: There is a real need to look at forecasting in the round. That means looking at the data, the method of forecasting, and the risk assessment and build this into departmental plans. We have a process map that acts as our mantra, a blueprint for people to check out what they have in place, where it works well, and where it isnt working as well as it could. This allows companies to focus attention on what is needed to improve forecasts. There is now the emergence of forecasters as leaders. Forecasters need to be able to say to senior managers, Im sorry, I cant model that. This is where there is scope to apply judgment, but not formally include these changes in the model.
How can forecasts keep up with changing markets?
Mete Sylan: There is scope to undertake research that will support forecasters. This includes working with physicians to understand some of the basic factors: patient flows, needs, and resource use.
Lauren Franzel: Our methodology and assumptions are approved and endorsed by all members of the sub-team that supports this work. This is important to provide credibility to forecasts that are used by a variety of stakeholders, including the pharmaceutical industry, to inform a host of decisions. Industry is an important consumer of our forecasts; they aid them in both short- and long-term decision-making, influences capacity planning, and helps provide insight into what specific countries are doing in the next one to five years but also much further in the future, up to 2030.
Nic Talbot Watt: We need to ensure that forecasts capture some of the key changes, like drug resistance and targeted therapies that make use of biological markers, which change the target population. So forecasts now need to focus on at-risk populations rather than just core prevalent groups. And this is exactly where dynamic approaches to forecasting can help.
Paul Goodwin: Sometimes, people make further judgmental adjustments because they think they have spotted a trend in the recent data. That can help update forecasts for changes in the market. But sometimes this is falsely reading patterns, and its best to leave the forecast alone. The trick is to know when to apply judgmental adjustments and when not to.