Certainity of fatigue life due to vibration cyclic loading ?

Recently a 50 year old bridge over railway track has fallen down in Kolkata of India.

People are raising doubt with what certainty we can predict fatigue life of any structure subjected to wear and tear, live load of vehicles over bridge and heavy vibration caused by a recently erected metro rail foundation structure.

For example, if we have piping subjected to vibration or cyclic thermal load, we can't predict the fatigue life with complete certainity. With some percentage of probability, we can produce a fatigue curve of a thick band instead of a line curve.

For a person of non STEM background, such justification can look strange.
But, what our forum members think - can we predict a timing of earthquake or a fatigue failure with complete certainity ?
reg,
Sam