In the present study, the maximum spreading diameter of a droplet impacting with a spherical particle is numerically studied for a wide range of impact conditions: Weber number (We) 0-110, Ohnesorge number (Oh) 0.0013-0.7869, equilibrium contact angle ( θeqi) 20{degree sign}-160{degree sign}, and droplet-to-particle size ratio (Ω) 1/10-1/2. A total of 2600 collision cases are simulated to enable a systematic analysis and prepare a large dataset for training of a data-driven prediction model. The effects of four impact parameters (We, Oh, θeqi, and Ω) on the maximum spreading diameter ( β*max) are comprehensively analyzed, and particular attention is paid to the difference of β*max between the low and high Weber number regimes. A universal model for prediction of β*max, as a function of We, Oh, θeqi, and Ω, is also proposed based on a deep neural network. It is shown that our data-driven model can predict the maximum spreading diameter well, showing an excellent agreement with the existing experimental results as well as our simulation dataset within a deviation range of {plus minus} 10%.