We　propose　a　machine　learning　embedded　method　of　parameters　determination　in　the　constitutional　models　of　hydrogels.　It　is　found　that　the　developed　logistic　regression-like　algorithm　for　hydrogel　swelling　allows　us　to　determine　the　fitting　parameters　based　on　known　swelling　ratio　and　chemical　potential.　We　also　put　forward　the　neural　networks-like　algorithm,　which,　by　its　own　property,　can　converge　faster　as　the　layer　deepens.　We　then　develop　neural　networks-like　algorithm　for　hydrogel　under　uniaxial　load　for　experimental　application　purpose.　Finally,　we　propose　several　machine　learning　embedded　potential　applications　for　hydrogels,　which　would　provide　directions　for　machine　learning-based　hydrogel　research.