A search for neutral Higgsino particles in final states with large missing transverse momentum and either a lepton-lepton pair or lepton-track pair is presented. The signal scenario considers neutralinos comprising two mass eigenstates differing in mass by small values of approximately 1-5 $/GeV$, where the heavier neutralino decays into the lighter neutralino and two same-flavor leptons. The leptons possess small transverse momentum and thus often fail to be reconstructed. To recover sensitivity, events in which only one of the leptons is identified, while the second lepton is measured as a simple track with the opposite charge, are considered, complementing the case where both leptons are reconstructed. A dedicated isolation method is used, that optimally selects signal leptons, which are often nearly collinear and spoil each other's standard isolation. The custom isolation variable serves to define control regions used to estimate standard model backgrounds for the search. Multivariate discriminants are used to enhance sensitivity at various stages of the selection. The search does not overlap with previous searches in the dilepton final state and probes signal model phase space that has not been explored by previous searches. The search is designed to analyze the proton-proton collision data collected with the CMS experiment during Run 2 with luminosity of 137 1/fb at a center-of-mass energy of 13 TeV. The interpretation is done with a simplified model of compressed mass Higgsinos. The full luminosity is used to calculate the background yield from data and the expected limits. Near the LEP limits of 100 GeV, mass difference between the chargino and neutralino down to 0.8 GeV is expected to be excluded. At higher mass splittings between 2 and 2.5 GeV, a chargino mass of up to almost 160 GeV is expected to be excluded. The expected limits improve upon previous searches with similar final states. For the purpose of this thesis, 10% of the data was unblinded showing good agreement between the measured data and the predicted background from Standard Model processes.